2_Two Layer Neural Network.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1 style=\"color:navy\">II. Two-Layer Neural Network</h1>\n",
    "\n",
    "İki katmanlı yapay sinir ağı, diğer çok katmanlı sinir ağları gibi temel olarak üç bölümden oluşur. **Input-Layer** (Giriş katmanı), **Hidden-Layer** (Gizli katman) ve **Output-Layer** (Çıkış katmanı). Katman sayısı belirlenirken, input harici tüm katmanlar sayılır. İki katmanlı yapay sinir ağı (YSA), Lojistik Regresyonda kullandığımız en basit sinir ağına, bir katman daha (hidden layer) eklenerek oluşturulmuştur. Hidden ve Output katmanlarındaki nöron sayıları, ağdan ağa farklılık gösterebilir. İki katmanlı YSA, lojistik regresyonun çözmekte yetersiz kaldığı problemlere daha iyi çözümler getirebilir. Katman sayısı arttıkça genelde ağın problem çözme yetisi de artmaktadır. Fakat bu her zaman doğru değildir. Bir probleme daha az katmanla, daha iyi çözümler getiren sinir ağları da bulunabilir."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"https://raw.githubusercontent.com/zkcplk/Deep-Learning/main/images/7_two_layer_neural_network.png\" width=\"800px\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lojistik Regresyonda kullandığımız tek aktivasyon fonksiyonu sigmoid idi. Burada ise gizli katmanda **hiperbolik tanjant (tanh)** fonksiyonu kullanılır. Bunun sebebi, neredeyse her zaman sigmoid fonksiyonundan daha iyi çalışmasıdır. Sigmoid fonksiyonu 0-1 aralığında değer alırken, Tanh fonksiyonu -1 ile +1 aralığında değer alır. Bazen bir derin öğrenme algoritmasını eğitmek istediğinizde, verilerinizi merkeze almak isteyebilirsiniz. Verileri merkeze almanın etkisi, verilerinizin ortalamasının 0.05 gibi veya 0'a yakın bir değer almasıdır. Bu da bir sonraki katman için öğrenmeyi kolaylaştırır. Fakat çıkış katmanı için hala sigmoid kullanmaya devam edebiliriz. Özellikle çıktının 0 veya 1 olmasını istediğimiz ikili sınıflandırma problemimiz varsa, mutlaka çıkış katmanında sigmoid fonksiyonunu kullanmamız gerekir. Sigmoid ve Tanh dışında daha derin sinir ağlarında, hem hesaplama hızını arttırmak hem de başka faydaları sebebiyle ReLU ve Leaky ReLU isimli aktivasyon fonksiyonları da kullanılır. Şimdilik bunlara değinmiyoruz."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"https://raw.githubusercontent.com/zkcplk/Deep-Learning/main/images/8_sigmoid_tanh_functions.jpg\" width=\"800px\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Katmanları sayısal olarak ifade ederken köşeli parantez kullanırız. **[1]** ifadesi birinci katmanı belirtir. Yukarıdaki örnekte gizli katman, aynı zamanda birinci katmandır. Birinci katmana ait $a$  değerleri, katmanda kullanılan **nöron sayısına göre değişkenlik göstermekle birlikte** şu şekilde sıralanır: \n",
    "\n",
    "$$a_1^{[1]} , a_2^{[1]} ,  a_3^{[1]} ...$$\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bu örnekte **gizli katmandaki her bir $a$ değeri** aslında bir tanh fonksiyonunun çıktısıdır: $a_{1,2,3,4}^{[1]}$ = $g$($z_{1,2,3,4}^{[1]}$) = $tanh$($z_{1,2,3,4}^{[1]}$)\n",
    "\n",
    "Çıktı katmanında ise: $a_{1}^{[2]}$ = $g$($z_{1}^{[2]}$) = $sigmoid$($z_{1}^{[2]}$)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$a_1^{[1]}$ $a_2^{[1]}$ $a_3^{[1]}$ $a_4^{[1]}$ ... bir vektör matrisi oluşturur ve $a^{[1]}$ şeklinde ifade edilir. $a^{[1]}$ gibi $z^{[1]}$ ve $b^{[1]}$ matrisleri de aynı mantıkla yazılabilir."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "$$a^{[1]} = \n",
    "\\begin{bmatrix}\n",
    "    a_1^{[1]}\\\\\n",
    "    a_2^{[1]}\\\\\n",
    "    \\vdots\\\\\n",
    "    a_4^{[1]}\n",
    "\\end{bmatrix} \\text{    } z^{[1]} = \n",
    "\\begin{bmatrix}\n",
    "    z_1^{[1]}\\\\\n",
    "    z_2^{[1]}\\\\\n",
    "    \\vdots\\\\\n",
    "    z_4^{[1]}\n",
    "\\end{bmatrix} \\text{    } b^{[1]} = \n",
    "\\begin{bmatrix}\n",
    "    b_1^{[1]}\\\\\n",
    "    b_2^{[1]}\\\\\n",
    "    \\vdots\\\\\n",
    "    b_4^{[1]}\n",
    "\\end{bmatrix}$$\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bir katmandaki her bir nöronda, o katmana giriş yapan x değerleri (değişkenleri) kadar $w$ değerleri bulunur. Örneğin, üstte verilen YSA incelenirse, 1. katmandaki her bir nöronda, o nörona giren $x_1$ ve $x_2$'den dolayı, $w_1$ ve $w_2$ değerleri vardır ve 1.  katmanda toplam 4 nöron olduğu için, $W^{[1]}$ matrisi (4x2)'lik bir matris olur.\n",
    "\n",
    "\n",
    "$$W^{[1]} = \n",
    "\\begin{bmatrix}\n",
    "    w_1^{[1][1]} w_1^{[1][2]}\\\\\n",
    "    w_2^{[1][1]} w_2^{[1][2]}\\\\\n",
    "    \\vdots\\\\\n",
    "    w_4^{[1][1]} w_4^{[1][2]}\n",
    "\\end{bmatrix} \\text{    }$$\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "Giriş katmanındaki $x_1$ $x_2$ $x_3$ ... değerleri ile bir vektör matrisi oluşturulursa, o da $a^{[0]}$ olarak ifade edilebilir. Aşağıda, her katmanda hesaplanan değerlerin matris boyutları verilmiştir.\n",
    "\n",
    "<br>\n",
    "$$a^{[1]} : (4,1)$$\n",
    "<br>\n",
    "$$z^{[1]} : (4,1)$$\n",
    "<br>\n",
    "$$b^{[1]} : (4,1)$$\n",
    "<br>\n",
    "$$W^{[1]} : (4,2)$$\n",
    "<br>\n",
    "$$a^{[2]} : (1,1)$$\n",
    "<br>\n",
    "$$z^{[2]} : (1,1)$$\n",
    "<br>\n",
    "$$b^{[2]} : (1,1)$$\n",
    "<br>\n",
    "$$W^{[2]} : (1,4)$$\n",
    "<br>\n",
    "\n",
    "\n",
    "Aşağıdaki resimde daha önce işlediğimiz Lojistik Regresyon ve şuan incelediğimiz İki katmanlı YSA'da gerçekleştirilen forward ve backward propagation işlemleri, karşılaştırmalı olarak gösterilmiştir.\n",
    "\n",
    "<img src=\"https://raw.githubusercontent.com/zkcplk/Deep-Learning/main/images/9_logistic_regression_two_layer_neural_network.png\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Daha önce lojistik regresyonda backward propagation yaparken, kullandığımız türevleme işlemleri, iki katmanlı ve daha sonra işleyeceğimiz çok katmanlı sinir ağlarında da geçerlidir. Türev almaktaki amaç, yeni $w$ ve $b$ değerleri üretmektir. Burada iki katman olduğu için her katmanda ayrı ayrı $w$ ve $b$ değerleri üretilecektir. Dolayısıyla kodumuzda, iki adet **dw** ve iki adet **db** türev değişkeni bulunacaktır.\n",
    "\n",
    "Yukarıdaki resimde her bir katman bir sonraki katmana bilgi taşımaktadır. Örneğin, birinci katmanda elde edilen $a^{[1]}$ değeri, ikinci katmanda $z^{[2]}$ denkleminde kullanılmıştır. Eğer örneğimizde bir katman daha olsaydı, ikinci katmandan üretilen $a^{[2]}$ değeri, bir sonraki katmanda $z^{[3]}$ denkleminde kullanılacaktı. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Yazacağımız kodda kullanacağımız formüller aşağıdaki gibidir. Önceki derste ifadelerle ilgili ayrıntılı açıklama yapıldığı için burada sadece formülleri vererek geçeceğiz. Örneğimizdeki YSA'da, veri setindeki her bir veri (i) için **z**, **a** ve **Cost** değeri aşağıdaki gibi hesaplanır:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "<!--\n",
    "$$z^{[1] (i)} =  W^{[1]} x^{(i)} + b^{[1]}\\tag{1}$$ \n",
    "    \n",
    "$$a^{[1] (i)} = tanh(z^{[1] (i)})\\tag{2}$$\n",
    "    \n",
    "$$z^{[2] (i)} = W^{[2]} a^{[1] (i)} + b^{[2]}\\tag{3}$$\n",
    "    \n",
    "$$\\hat{y}^{(i)} = a^{[2] (i)} = sigmoid(z^{ [2] (i)})\\tag{4}$$\n",
    "    \n",
    "$$ \\mathcal{L}(a^{[2] (i)}, y^{(i)}) =  - y^{(i)}  \\log(a^{[2] (i)}) + (1-y^{(i)} )  \\log(1-a^{[2] (i)}) \\tag{5}$$\n",
    "\n",
    "$$J = - \\frac{1}{m} \\sum\\limits_{i = 0}^{m} \\mathcal{L}(a^{[2] (i)}, y^{(i)}) \\tag{6}$$\n",
    "-->\n",
    "\n",
    "\n",
    "<img src=\"https://raw.githubusercontent.com/zkcplk/Deep-Learning/main/images/formul1.png\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Şimdi de backward propagation'da işimize yarayacak diğer ifadeleri yazalım. Bu ifadeler, yukarıdaki şekilde anlatılan türev işlemlerinden sonra elde edilmiştir. Matematiksel ayrıntı konumuz dışında olduğu ve uzun süreceği için direkt formüller üzerinden ilerleyeceğiz. Kod içerisinde hangi formülün nerede kullanıldığı belirtilecektir.\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--\n",
    "$$ dZ^{[2]} = A^{[2]} - Y \\tag{7}$$\n",
    "\n",
    "$$ dW^{[2]} = \\frac{1}{m} dZ^{[2]} A^{[1]^{T}} \\tag{8}$$\n",
    "    \n",
    "$$ db^{[2]} = \\frac{1}{m} np.sum(dZ^{[2]},axis=1,keepdims=True) \\tag{9}$$\n",
    "    \n",
    "$$ dZ^{[1]} = W^{[2]^{T}} dZ^{[2]} * g^{[1]^{`}}(Z^{[1]}) \\tag{10}$$\n",
    "\n",
    "$$ dW^{[1]} = \\frac{1}{m} dZ^{[1]} X^{T} \\tag{11}$$\n",
    "\n",
    "$$ db^{[1]} = \\frac{1}{m} np.sum(dZ^{[1]},axis=1,keepdims=True) \\tag{12}$$\n",
    "-->\n",
    "\n",
    "\n",
    "<img src=\"https://raw.githubusercontent.com/zkcplk/Deep-Learning/main/images/formul2.png\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2 katmanlı YSA'nın teorisine değindikten sonra, kodlamalarımızda kullanacağımız fonksiyonları oluşturalım."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# numpy & pandas.\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "#====================\n",
    "# sigmoid fonksiyonu.\n",
    "def sigmoid(z):\n",
    "    return 1/(1+np.exp(-z))\n",
    "\n",
    "#====================\n",
    "# başlangıç parametrelerini belirlemek.\n",
    "# iki katmanlı bir YSA olduğu için hidden layer, 1 katmandan oluşuyor.\n",
    "# hidden değeri gizli katmandaki nöron sayısını ifade ediyor.\n",
    "def initialize_parameters(X, Y, hidden):\n",
    "    # input layer\n",
    "    n_x = X.shape[0]\n",
    "    \n",
    "    # hidden layer\n",
    "    n_h = hidden\n",
    "    \n",
    "    # output layer\n",
    "    n_y = Y.shape[0]\n",
    "    \n",
    "    # başka kişilerle / başka zamanda aynı sonuçları bulabilmek için aynı random ayarını yapmalıyız.\n",
    "    np.random.seed(2) \n",
    "    \n",
    "    # lojistik regresyonda başlangıç w ve b değerlerini 0 verebiliyorduk.\n",
    "    # fakat çok katmanlı YSA için bu geçerli değildir, sıfırdan farklı bir değer vermek gerekir.\n",
    "    # nöron sayısı arttıkça bulunacak random sayıların değeri de büyüyebilecektir.\n",
    "    # o yüzden 0.01 ile çarparak, çıkacak olan değeri iyice küçültmüş oluyoruz.\n",
    "    W1 = np.random.randn(n_h, n_x) * 0.01\n",
    "    b1 = np.zeros(shape=(n_h, 1))\n",
    "    W2 = np.random.randn(n_y, n_h) * 0.01\n",
    "    b2 = np.zeros(shape=(n_y, 1))\n",
    "    \n",
    "    # sonuç\n",
    "    parameters = {\"W1\": W1,\n",
    "                  \"b1\": b1,\n",
    "                  \"W2\": W2,\n",
    "                  \"b2\": b2}\n",
    "    \n",
    "    return parameters\n",
    "\n",
    "#====================\n",
    "# yine lojistik regresyondan farklı olarak \n",
    "# forward ve backward propagation işlemlerini ayrı ayrı gerçekleştireceğiz.\n",
    "# iki katmanlı bir YSA olduğu için forward propagation'da A1, Z1, A2 ve Z2 değerlerini hesaplıyoruz.\n",
    "# backward prop.'da yeniden hesaplama yapmamak ve hazır kullanmak için cache alıyoruz.\n",
    "def forward_propagation(X, parameters):\n",
    "    # initialize_parameters() fonksiyonundan gelen değerler.\n",
    "    W1 = parameters['W1']\n",
    "    b1 = parameters['b1']\n",
    "    W2 = parameters['W2']\n",
    "    b2 = parameters['b2']\n",
    "    \n",
    "    # (1) numaralı formül.\n",
    "    Z1 = np.dot(W1, X) + b1\n",
    "    \n",
    "    # (2) numaralı formül.\n",
    "    A1 = np.tanh(Z1)\n",
    "    \n",
    "    # (3) numaralı formül.\n",
    "    Z2 = np.dot(W2, A1) + b2\n",
    "    \n",
    "    # (4) numaralı formül.\n",
    "    A2 = sigmoid(Z2)\n",
    "    \n",
    "    cache = {\"Z1\": Z1,\n",
    "             \"A1\": A1,\n",
    "             \"Z2\": Z2,\n",
    "             \"A2\": A2}\n",
    "    \n",
    "    return A2, cache\n",
    "\n",
    "#====================\n",
    "# cost değerinin hesaplanması.\n",
    "def compute_cost(A2, Y):\n",
    "    # veri miktarı.\n",
    "    m = Y.shape[1] \n",
    "    \n",
    "    # (5) ve (6) numaralı formüller.\n",
    "    logprobs = np.multiply(np.log(A2), Y) + np.multiply((1 - Y), np.log(1 - A2))\n",
    "    cost = - np.sum(logprobs) / m\n",
    "    \n",
    "    # vektörel sonucu skaler sonuca döndürmek için.\n",
    "    # örnek: [[17]] --> 17 \n",
    "    cost = float(np.squeeze(cost))  \n",
    "    \n",
    "    # sonuç\n",
    "    return cost\n",
    "\n",
    "#====================\n",
    "# backward propagation\n",
    "def backward_propagation(parameters, cache, X, Y):\n",
    "    # veri miktarı.\n",
    "    m = X.shape[1]\n",
    "    \n",
    "    # daha önce hesaplanan değerlerin cache'ten çekilmesi.\n",
    "    W1 = parameters['W1']\n",
    "    W2 = parameters['W2']    \n",
    "    A1 = cache['A1']\n",
    "    A2 = cache['A2']\n",
    "\n",
    "    # türevlerin hesaplanması.\n",
    "    # (7) numaralı formül.\n",
    "    dZ2 = A2 - Y\n",
    "    \n",
    "    # (8) numaralı formül.\n",
    "    dW2 = (1 / m) * np.dot(dZ2, A1.T)\n",
    "    \n",
    "    # (9) numaralı formül.\n",
    "    db2 = (1 / m) * np.sum(dZ2, axis=1, keepdims=True)\n",
    "    \n",
    "    # (10) numaralı formül.\n",
    "    dZ1 = np.multiply(np.dot(W2.T, dZ2), 1 - np.power(A1, 2))\n",
    "    \n",
    "    # (11) numaralı formül.\n",
    "    dW1 = (1 / m) * np.dot(dZ1, X.T)\n",
    "    \n",
    "    # (12) numaralı formül.\n",
    "    db1 = (1 / m) * np.sum(dZ1, axis=1, keepdims=True)\n",
    "    \n",
    "    # sonuç\n",
    "    grads = {\"dW1\": dW1,\n",
    "             \"db1\": db1,\n",
    "             \"dW2\": dW2,\n",
    "             \"db2\": db2}\n",
    "    \n",
    "    return grads\n",
    "\n",
    "#====================\n",
    "# gradient descent'ın uygulanması.\n",
    "def update_parameters(parameters, grads, learning_rate):\n",
    "    # daha önce hesaplanan değerlerin çekilmesi.\n",
    "    W1 = parameters['W1']\n",
    "    b1 = parameters['b1']\n",
    "    W2 = parameters['W2']\n",
    "    b2 = parameters['b2']\n",
    "    dW1 = grads['dW1']\n",
    "    db1 = grads['db1']\n",
    "    dW2 = grads['dW2']\n",
    "    db2 = grads['db2']\n",
    "\n",
    "    # yeni w ve b değerlerinin üretilmesi.\n",
    "    W1 = W1 - learning_rate * dW1\n",
    "    b1 = b1 - learning_rate * db1\n",
    "    W2 = W2 - learning_rate * dW2\n",
    "    b2 = b2 - learning_rate * db2\n",
    "    \n",
    "    # sonuç.\n",
    "    parameters = {\"W1\": W1,\n",
    "                  \"b1\": b1,\n",
    "                  \"W2\": W2,\n",
    "                  \"b2\": b2}\n",
    "    \n",
    "    return parameters\n",
    "\n",
    "#====================\n",
    "# oluşturduğumuz tüm fonksiyonları bir model fonksiyonunda bir araya getirelim.\n",
    "def nn_model(X_train, Y_train, X_test, Y_test, hidden = 4, epoch = 3000, learning_rate = 1.2, print_cost=False):\n",
    "    costs = []\n",
    "    \n",
    "    print(\"X_train boyutu: \", X_train.shape)\n",
    "    print(\"Y_train boyutu: \", Y_train.shape)\n",
    "    print(\"X_test boyutu: \", X_test.shape)\n",
    "    print(\"Y_test boyutu: \", Y_test.shape)\n",
    "    \n",
    "    # başlangıç parametreleri belirleniyor...\n",
    "    parameters = initialize_parameters(X_train, Y_train, hidden)\n",
    "\n",
    "    # girilen dönem (epoch) sayısı kadar hesaplama yapılacak.\n",
    "    for i in range(0, epoch):\n",
    "        \n",
    "        # forward propagation.\n",
    "        A2, cache = forward_propagation(X_train, parameters)\n",
    "        \n",
    "        # cost hesaplanıyor.\n",
    "        cost = compute_cost(A2, Y_train)\n",
    " \n",
    "        # backward propagation.\n",
    "        grads = backward_propagation(parameters, cache, X_train, Y_train)\n",
    " \n",
    "        # gradient descent\n",
    "        # yeni w ve b değerlerinin üretilmesi.\n",
    "        parameters = update_parameters(parameters, grads, learning_rate)\n",
    "        \n",
    "        # 100 dönemde bir cost kaydediliyor, böylece ne kadar azaldığını grafikten izleyebileceğiz.\n",
    "        if i % 100 == 0:\n",
    "            costs.append(cost)\n",
    "            \n",
    "        # 100 dönemde bir cost değerinin ekrana basılması.\n",
    "        if print_cost and i % 100 == 0:\n",
    "            print (\"Epoch: %i, Cost: %f\" %(i, cost))\n",
    "\n",
    "    # train ve test veri setleri ile tüm tahminler hesaplanıyor.\n",
    "    Y_hat_test = predict(parameters, X_test)\n",
    "    Y_hat_train = predict(parameters, X_train)\n",
    "\n",
    "    # yapılan tahminler ile gerçek veriler karşılaştırılıp, doğruluk oranı hesaplanıyor.\n",
    "    print(\"Train Accuracy: {} %\".format(100 - np.mean(np.abs(Y_hat_train - Y_train)) * 100))\n",
    "    print(\"Test Accuracy: {} %\".format(100 - np.mean(np.abs(Y_hat_test - Y_test)) * 100))\n",
    "    \n",
    "    # bulunan tüm sonuçlar bir sözlükte toplanıyor.\n",
    "    d = {\"costs\": costs,\n",
    "         \"Y_hat_test\": Y_hat_test, \n",
    "         \"Y_hat_train\" : Y_hat_train, \n",
    "         \"parameters\" : parameters, \n",
    "         \"learning_rate\" : learning_rate,\n",
    "         \"epoch\": epoch}\n",
    "    \n",
    "    return d\n",
    "\n",
    "#====================\n",
    "# tahmin işlemi.\n",
    "def predict(parameters, X):\n",
    "    # tahmin edilecek y değerleri (y şapka).\n",
    "    Y_hat = np.zeros((1,X.shape[1]))\n",
    "    \n",
    "    A2, cache = forward_propagation(X, parameters)\n",
    "    pred = np.round(A2)\n",
    "    \n",
    "    for i in range(pred.shape[1]):\n",
    "        if (A2[0,i] > 0.5): \n",
    "            Y_hat[0,i] = 1.\n",
    "        else:\n",
    "            Y_hat[0,i] = 0.\n",
    "    \n",
    "    return Y_hat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Şimdi kullanabileceğimiz fonksiyonlar hazır olduğuna göre, örnek bir veri seti için 2 katmanlı YSA modelimizi çalıştırabiliriz.  Amaç sadece 2 katmanlı YSA'nın nasıl çalıştığını ve yukarıda anlatılan formüllerin kullanımını göstermek olduğundan, veri analizi ve veri setinin yorumlanması vb. ayrıntılara girilmemiştir. 2 katmanlı YSA ile lojistik regresyon sonuçlarını karşılaştırmak için, daha önceki derste kullandığımız verisetlerinin aynısını kullanacağız."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ÖRNEK 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train boyutu:  (27, 140)\n",
      "Y_train boyutu:  (1, 140)\n",
      "X_test boyutu:  (27, 60)\n",
      "Y_test boyutu:  (1, 60)\n",
      "Epoch: 0, Cost: 0.693187\n",
      "Epoch: 100, Cost: 0.665616\n",
      "Epoch: 200, Cost: 0.645817\n",
      "Epoch: 300, Cost: 0.631044\n",
      "Epoch: 400, Cost: 0.619036\n",
      "Epoch: 500, Cost: 0.607645\n",
      "Epoch: 600, Cost: 0.594654\n",
      "Epoch: 700, Cost: 0.578142\n",
      "Epoch: 800, Cost: 0.557588\n",
      "Epoch: 900, Cost: 0.534757\n",
      "Epoch: 1000, Cost: 0.512754\n",
      "Epoch: 1100, Cost: 0.493899\n",
      "Epoch: 1200, Cost: 0.478780\n",
      "Epoch: 1300, Cost: 0.466823\n",
      "Epoch: 1400, Cost: 0.457158\n",
      "Epoch: 1500, Cost: 0.449059\n",
      "Epoch: 1600, Cost: 0.442034\n",
      "Epoch: 1700, Cost: 0.435776\n",
      "Epoch: 1800, Cost: 0.430100\n",
      "Epoch: 1900, Cost: 0.424884\n",
      "Epoch: 2000, Cost: 0.420045\n",
      "Epoch: 2100, Cost: 0.415520\n",
      "Epoch: 2200, Cost: 0.411258\n",
      "Epoch: 2300, Cost: 0.407219\n",
      "Epoch: 2400, Cost: 0.403367\n",
      "Epoch: 2500, Cost: 0.399674\n",
      "Epoch: 2600, Cost: 0.396119\n",
      "Epoch: 2700, Cost: 0.392686\n",
      "Epoch: 2800, Cost: 0.389362\n",
      "Epoch: 2900, Cost: 0.386140\n",
      "Epoch: 3000, Cost: 0.383013\n",
      "Epoch: 3100, Cost: 0.379980\n",
      "Epoch: 3200, Cost: 0.377038\n",
      "Epoch: 3300, Cost: 0.374188\n",
      "Epoch: 3400, Cost: 0.371429\n",
      "Epoch: 3500, Cost: 0.368762\n",
      "Epoch: 3600, Cost: 0.366187\n",
      "Epoch: 3700, Cost: 0.363703\n",
      "Epoch: 3800, Cost: 0.361310\n",
      "Epoch: 3900, Cost: 0.359008\n",
      "Epoch: 4000, Cost: 0.356795\n",
      "Epoch: 4100, Cost: 0.354670\n",
      "Epoch: 4200, Cost: 0.352631\n",
      "Epoch: 4300, Cost: 0.350675\n",
      "Epoch: 4400, Cost: 0.348801\n",
      "Epoch: 4500, Cost: 0.347006\n",
      "Epoch: 4600, Cost: 0.345286\n",
      "Epoch: 4700, Cost: 0.343639\n",
      "Epoch: 4800, Cost: 0.342061\n",
      "Epoch: 4900, Cost: 0.340550\n",
      "Epoch: 5000, Cost: 0.339101\n",
      "Epoch: 5100, Cost: 0.337713\n",
      "Epoch: 5200, Cost: 0.336380\n",
      "Epoch: 5300, Cost: 0.335102\n",
      "Epoch: 5400, Cost: 0.333873\n",
      "Epoch: 5500, Cost: 0.332692\n",
      "Epoch: 5600, Cost: 0.331556\n",
      "Epoch: 5700, Cost: 0.330462\n",
      "Epoch: 5800, Cost: 0.329407\n",
      "Epoch: 5900, Cost: 0.328389\n",
      "Epoch: 6000, Cost: 0.327406\n",
      "Epoch: 6100, Cost: 0.326455\n",
      "Epoch: 6200, Cost: 0.325534\n",
      "Epoch: 6300, Cost: 0.324642\n",
      "Epoch: 6400, Cost: 0.323775\n",
      "Epoch: 6500, Cost: 0.322933\n",
      "Epoch: 6600, Cost: 0.322114\n",
      "Epoch: 6700, Cost: 0.321315\n",
      "Epoch: 6800, Cost: 0.320535\n",
      "Epoch: 6900, Cost: 0.319772\n",
      "Epoch: 7000, Cost: 0.319025\n",
      "Epoch: 7100, Cost: 0.318292\n",
      "Epoch: 7200, Cost: 0.317572\n",
      "Epoch: 7300, Cost: 0.316863\n",
      "Epoch: 7400, Cost: 0.316163\n",
      "Epoch: 7500, Cost: 0.315472\n",
      "Epoch: 7600, Cost: 0.314788\n",
      "Epoch: 7700, Cost: 0.314109\n",
      "Epoch: 7800, Cost: 0.313434\n",
      "Epoch: 7900, Cost: 0.312761\n",
      "Epoch: 8000, Cost: 0.312090\n",
      "Epoch: 8100, Cost: 0.311420\n",
      "Epoch: 8200, Cost: 0.310748\n",
      "Epoch: 8300, Cost: 0.310074\n",
      "Epoch: 8400, Cost: 0.309397\n",
      "Epoch: 8500, Cost: 0.308715\n",
      "Epoch: 8600, Cost: 0.308027\n",
      "Epoch: 8700, Cost: 0.307332\n",
      "Epoch: 8800, Cost: 0.306629\n",
      "Epoch: 8900, Cost: 0.305917\n",
      "Epoch: 9000, Cost: 0.305195\n",
      "Epoch: 9100, Cost: 0.304462\n",
      "Epoch: 9200, Cost: 0.303717\n",
      "Epoch: 9300, Cost: 0.302960\n",
      "Epoch: 9400, Cost: 0.302188\n",
      "Epoch: 9500, Cost: 0.301403\n",
      "Epoch: 9600, Cost: 0.300602\n",
      "Epoch: 9700, Cost: 0.299785\n",
      "Epoch: 9800, Cost: 0.298953\n",
      "Epoch: 9900, Cost: 0.298103\n",
      "Epoch: 10000, Cost: 0.297237\n",
      "Epoch: 10100, Cost: 0.296353\n",
      "Epoch: 10200, Cost: 0.295451\n",
      "Epoch: 10300, Cost: 0.294530\n",
      "Epoch: 10400, Cost: 0.293592\n",
      "Epoch: 10500, Cost: 0.292634\n",
      "Epoch: 10600, Cost: 0.291658\n",
      "Epoch: 10700, Cost: 0.290662\n",
      "Epoch: 10800, Cost: 0.289648\n",
      "Epoch: 10900, Cost: 0.288614\n",
      "Epoch: 11000, Cost: 0.287560\n",
      "Epoch: 11100, Cost: 0.286487\n",
      "Epoch: 11200, Cost: 0.285395\n",
      "Epoch: 11300, Cost: 0.284282\n",
      "Epoch: 11400, Cost: 0.283151\n",
      "Epoch: 11500, Cost: 0.282000\n",
      "Epoch: 11600, Cost: 0.280830\n",
      "Epoch: 11700, Cost: 0.279641\n",
      "Epoch: 11800, Cost: 0.278433\n",
      "Epoch: 11900, Cost: 0.277206\n",
      "Train Accuracy: 88.57142857142857 %\n",
      "Test Accuracy: 75.0 %\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "churn_data = pd.read_csv(\"datasets/churn_data.csv\")\n",
    "df1 = churn_data.copy()\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X = df1.drop(['churn'], axis=1)\n",
    "Y = df1['churn']\n",
    "\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.3, random_state=30)\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "scaler = StandardScaler()\n",
    "scaler.fit(X_train)\n",
    "\n",
    "X_train_scaled = scaler.transform(X_train)\n",
    "X_test_scaled = scaler.transform(X_test)\n",
    "\n",
    "X_train_scaled = X_train_scaled.T\n",
    "X_test_scaled = X_test_scaled.T\n",
    "\n",
    "Y_train = Y_train.to_numpy().reshape(1,X_train_scaled.shape[1])\n",
    "Y_test = Y_test.to_numpy().reshape(1,X_test_scaled.shape[1])\n",
    "\n",
    "d1 = nn_model(X_train_scaled, Y_train, X_test_scaled, Y_test, \n",
    "              hidden = 11, epoch = 12000, learning_rate = 0.007, print_cost=True)\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.plot(d1['costs'])\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Daha önce aynı veri setini Lojistik Regresyon ile modellediğimizde;\n",
    "* Train Accuracy: 85.71 %\n",
    "* Test Accuracy: 73.33 %\n",
    "\n",
    "değerlerine ulaşmıştık. 2 katmanlı YSA modeli ise, aynı veri seti için daha iyi sonuçlar vermiş oldu."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'costs': [0.6931866192045918,\n",
       "  0.6656163288891687,\n",
       "  0.6458171573256349,\n",
       "  0.6310438709399435,\n",
       "  0.6190355338829092,\n",
       "  0.6076445012134604,\n",
       "  0.5946542411613716,\n",
       "  0.5781420708584766,\n",
       "  0.5575877335742648,\n",
       "  0.5347569320296125,\n",
       "  0.5127536072977337,\n",
       "  0.493899430430647,\n",
       "  0.4787798660371146,\n",
       "  0.4668229919794178,\n",
       "  0.4571581042515897,\n",
       "  0.4490588575005818,\n",
       "  0.44203350663644464,\n",
       "  0.43577641013895946,\n",
       "  0.4300999754965971,\n",
       "  0.4248839755020835,\n",
       "  0.4200449740184133,\n",
       "  0.41551997432428883,\n",
       "  0.41125844773344555,\n",
       "  0.4072187453113412,\n",
       "  0.40336653839955583,\n",
       "  0.3996740383542253,\n",
       "  0.396119404871694,\n",
       "  0.39268611271029596,\n",
       "  0.38936222570941725,\n",
       "  0.3861396044218496,\n",
       "  0.383013098949168,\n",
       "  0.379979779530964,\n",
       "  0.37703824778123285,\n",
       "  0.37418805716524767,\n",
       "  0.37142925534460886,\n",
       "  0.36876204590583167,\n",
       "  0.3661865552533475,\n",
       "  0.3637026838581819,\n",
       "  0.3613100198316217,\n",
       "  0.3590077956170731,\n",
       "  0.356794873370094,\n",
       "  0.3546697494217882,\n",
       "  0.35263057196282294,\n",
       "  0.35067516843753693,\n",
       "  0.3488010803306629,\n",
       "  0.34700560348448123,\n",
       "  0.34528583218038184,\n",
       "  0.34363870520423695,\n",
       "  0.3420610521176575,\n",
       "  0.34054963802976557,\n",
       "  0.33910120532327354,\n",
       "  0.3377125110352116,\n",
       "  0.336380358915867,\n",
       "  0.33510162556605777,\n",
       "  0.33387328044793196,\n",
       "  0.3326923999376878,\n",
       "  0.33155617590277503,\n",
       "  0.33046191951521037,\n",
       "  0.32940706114588664,\n",
       "  0.3283891472264907,\n",
       "  0.3274058349314218,\n",
       "  0.32645488544349716,\n",
       "  0.32553415644668104,\n",
       "  0.3246415943563347,\n",
       "  0.32377522666757647,\n",
       "  0.32293315468518785,\n",
       "  0.32211354679942444,\n",
       "  0.3213146323928821,\n",
       "  0.32053469640366167,\n",
       "  0.3197720745274827,\n",
       "  0.3190251490135501,\n",
       "  0.31829234499315046,\n",
       "  0.31757212727362044,\n",
       "  0.3168629975312598,\n",
       "  0.3161634918430749,\n",
       "  0.315472178507427,\n",
       "  0.3147876561164852,\n",
       "  0.3141085518579225,\n",
       "  0.31343352003876795,\n",
       "  0.3127612408401297,\n",
       "  0.3120904193270063,\n",
       "  0.31141978475191046,\n",
       "  0.31074809020361055,\n",
       "  0.3100741126617215,\n",
       "  0.30939665352255474,\n",
       "  0.30871453965968954,\n",
       "  0.3080266250721748,\n",
       "  0.30733179315237624,\n",
       "  0.30662895957322167,\n",
       "  0.30591707575121013,\n",
       "  0.30519513278903504,\n",
       "  0.30446216574410423,\n",
       "  0.30371725801273736,\n",
       "  0.30295954557206795,\n",
       "  0.3021882207909454,\n",
       "  0.3014025355149659,\n",
       "  0.3006018031544797,\n",
       "  0.29978539955989114,\n",
       "  0.2989527625533214,\n",
       "  0.29810339009302717,\n",
       "  0.2972368371664205,\n",
       "  0.29635271162644794,\n",
       "  0.2954506692910772,\n",
       "  0.2945304087044308,\n",
       "  0.2935916660008727,\n",
       "  0.2926342103139077,\n",
       "  0.29165784012802365,\n",
       "  0.29066238088598856,\n",
       "  0.2896476840431785,\n",
       "  0.2886136276146588,\n",
       "  0.2875601181033468,\n",
       "  0.2864870935440021,\n",
       "  0.2853945272639473,\n",
       "  0.28428243186216867,\n",
       "  0.2831508628557604,\n",
       "  0.28199992144386854,\n",
       "  0.2808297558956483,\n",
       "  0.27964056117487673,\n",
       "  0.27843257655807574,\n",
       "  0.27720608116885626],\n",
       " 'Y_hat_test': array([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0.,\n",
       "         0., 1., 1., 0., 1., 0., 0., 0., 0., 0., 0., 1., 1., 0., 0., 1.,\n",
       "         0., 1., 1., 1., 0., 1., 0., 1., 0., 0., 1., 1., 0., 0., 0., 1.,\n",
       "         0., 0., 0., 1., 0., 0., 1., 0., 0., 0., 0., 0.]]),\n",
       " 'Y_hat_train': array([[0., 0., 0., 0., 0., 0., 1., 1., 0., 0., 1., 0., 1., 0., 1., 0.,\n",
       "         0., 1., 0., 0., 0., 0., 1., 0., 0., 1., 1., 0., 0., 0., 0., 0.,\n",
       "         1., 1., 0., 0., 1., 0., 0., 0., 0., 0., 1., 0., 1., 0., 1., 0.,\n",
       "         0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0., 1.,\n",
       "         0., 0., 0., 0., 1., 1., 0., 1., 0., 0., 0., 0., 0., 1., 0., 1.,\n",
       "         1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0.,\n",
       "         0., 0., 1., 0., 0., 0., 0., 1., 1., 1., 0., 0., 0., 1., 0., 0.,\n",
       "         0., 1., 0., 0., 1., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0.,\n",
       "         0., 1., 1., 0., 0., 0., 1., 0., 0., 1., 0., 0.]]),\n",
       " 'parameters': {'W1': array([[-2.36254684e-01, -1.64509677e-01,  1.37427908e-01,\n",
       "          -4.32533310e-02, -9.40866164e-02, -1.27366820e-01,\n",
       "           4.12066431e-02, -1.67660813e-01, -1.52548940e-01,\n",
       "           9.55960565e-02,  1.97691613e-01, -1.33246318e-02,\n",
       "           1.86398002e-01, -7.88628691e-03,  5.51454135e-02,\n",
       "           2.35925775e-01, -1.37528158e-01,  4.59233104e-02,\n",
       "          -1.04605511e-01,  1.27213592e-02, -2.58937533e-02,\n",
       "           3.22317442e-02,  1.46562469e-01,  1.55931084e-01,\n",
       "           1.70572372e-02,  1.27369422e-01,  5.61623931e-02],\n",
       "         [-3.45227796e-01, -2.31521265e-01,  3.92113525e-02,\n",
       "          -5.44013152e-02,  3.07436197e-02, -2.26527071e-01,\n",
       "           1.04786428e-01, -2.81522021e-01, -1.60829148e-01,\n",
       "           6.96397283e-02,  1.16843844e-01, -7.12937630e-03,\n",
       "           2.43408947e-01,  5.23974221e-03,  2.70968150e-02,\n",
       "           1.62183347e-01, -1.16430191e-01,  8.10154737e-02,\n",
       "          -1.03193639e-01,  7.85908325e-02,  4.47258177e-02,\n",
       "          -1.07225369e-01,  1.91438722e-01,  1.16243688e-01,\n",
       "          -1.88755948e-02,  1.52394198e-01,  2.16406018e-02],\n",
       "         [ 2.03742574e-01,  1.27782202e-01, -1.94153104e-01,\n",
       "           4.79465447e-02,  4.91789870e-02,  6.27257173e-02,\n",
       "          -6.44520730e-02,  8.75776334e-02,  1.14503560e-01,\n",
       "          -1.06260177e-01, -2.32468998e-01,  1.42722989e-02,\n",
       "          -2.03953434e-01, -2.71521199e-02, -5.28328471e-02,\n",
       "          -2.31740310e-01,  1.02093512e-01, -2.80097416e-02,\n",
       "           6.89349392e-02,  5.49248484e-02, -1.04992690e-02,\n",
       "          -1.06096036e-02, -1.77036656e-01, -1.60617365e-01,\n",
       "          -1.07005451e-02, -1.54788519e-01, -1.06014895e-01],\n",
       "         [-1.64619334e-02, -5.71375105e-03,  2.57118230e-02,\n",
       "          -9.59575001e-03, -1.84314562e-02,  5.64107242e-03,\n",
       "           1.17708453e-02, -6.71774182e-03, -4.48729821e-03,\n",
       "           1.54098694e-02,  2.75069521e-02, -7.72885628e-04,\n",
       "           2.93925937e-02, -1.15533341e-02,  1.61484881e-02,\n",
       "           2.90970085e-02, -1.56682709e-02,  1.19000608e-02,\n",
       "          -3.26637080e-02,  1.07250313e-02,  4.07181954e-03,\n",
       "           8.08980343e-05,  2.46959505e-02,  1.72435067e-02,\n",
       "           3.20625528e-02,  3.05142304e-02,  1.08928111e-02],\n",
       "         [-3.16888095e-01, -1.97225964e-01,  1.46088236e-01,\n",
       "          -7.15010125e-02, -7.94428107e-02, -1.70117692e-01,\n",
       "           7.22371346e-02, -2.34927457e-01, -1.65097044e-01,\n",
       "           8.96501316e-02,  1.90175143e-01, -3.15390545e-02,\n",
       "           2.00395483e-01,  1.24024775e-02,  3.48859997e-02,\n",
       "           2.19998570e-01, -1.51639826e-01,  7.72872268e-02,\n",
       "          -8.89452307e-02,  2.77903828e-02, -3.15096549e-02,\n",
       "          -4.23501726e-02,  1.59816899e-01,  1.55453613e-01,\n",
       "          -5.50582300e-02,  1.69900880e-01,  6.21494528e-02],\n",
       "         [ 3.16893221e-01,  2.20998587e-01, -3.11351308e-04,\n",
       "           7.98750939e-02,  1.35028591e-02,  2.61679891e-01,\n",
       "          -1.71707941e-01,  4.55613536e-01,  1.32906843e-01,\n",
       "          -4.20006096e-02, -8.91810311e-02, -2.65124460e-02,\n",
       "          -2.25723496e-01, -2.05769194e-02, -2.62584782e-02,\n",
       "          -1.63160327e-01,  1.02782461e-01, -8.33942466e-02,\n",
       "          -5.05382629e-02, -1.49965566e-01,  5.20834395e-02,\n",
       "           1.98025516e-01, -5.03396174e-02, -8.00894540e-02,\n",
       "           8.61754093e-02, -1.13349490e-01, -1.06864309e-01],\n",
       "         [-3.57959084e-01, -1.91558958e-01,  9.51832497e-02,\n",
       "          -9.21692140e-02, -1.51122011e-01, -2.68445087e-01,\n",
       "           1.81541588e-01, -5.39063013e-01, -1.18754841e-01,\n",
       "           1.96315591e-02,  1.36569883e-01,  2.50437745e-02,\n",
       "           2.53541727e-01,  6.06522878e-02,  4.17368215e-03,\n",
       "           1.91330671e-01, -9.78124079e-02,  1.48529970e-01,\n",
       "           1.32517107e-02,  2.04566850e-01, -1.31586382e-01,\n",
       "          -1.70565325e-01, -9.99636689e-02,  8.08721881e-02,\n",
       "          -1.72151793e-01,  6.73196223e-02,  2.12293354e-01],\n",
       "         [-3.45765292e-02, -2.57942790e-02,  3.96516622e-02,\n",
       "          -2.30946967e-02, -1.68632570e-02, -3.78054874e-02,\n",
       "           2.43577338e-02, -2.04311671e-02, -1.32835310e-02,\n",
       "           3.45911642e-02,  5.00785696e-02, -1.48792823e-02,\n",
       "           2.83070061e-02,  1.18567375e-02,  2.62423547e-02,\n",
       "           7.06555418e-02, -3.97778455e-02,  1.38777991e-02,\n",
       "          -1.27672039e-02, -7.78926819e-03, -8.65290622e-04,\n",
       "           4.19135932e-04,  4.84790021e-02,  3.81816126e-02,\n",
       "           2.61092751e-03,  1.95798611e-02,  1.96527048e-02],\n",
       "         [ 2.16123385e-01,  1.43884022e-01, -1.35553033e-01,\n",
       "           4.29542594e-02,  5.66907737e-02,  1.27047643e-01,\n",
       "          -4.18078708e-02,  1.38967435e-01,  1.26713606e-01,\n",
       "          -9.19220442e-02, -1.57221870e-01,  2.63629592e-02,\n",
       "          -1.56156386e-01,  2.47248680e-03, -2.38125116e-02,\n",
       "          -1.97540727e-01,  1.23820170e-01, -2.53870450e-02,\n",
       "           7.92700755e-02, -2.24134739e-03, -8.97324284e-03,\n",
       "          -5.37970721e-03, -1.51004000e-01, -1.41697453e-01,\n",
       "          -3.09282035e-02, -1.36109670e-01, -4.37016596e-02],\n",
       "         [ 4.27605381e-01,  2.24394648e-01, -1.07103103e-03,\n",
       "           9.64875279e-02,  1.24614719e-01,  3.24173156e-01,\n",
       "          -2.29950400e-01,  6.84244747e-01,  1.48177751e-01,\n",
       "           1.71415727e-02, -1.02032861e-01, -7.30003537e-02,\n",
       "          -3.70621766e-01, -9.61848503e-02,  1.03746570e-02,\n",
       "          -1.51905223e-01,  4.12306517e-02, -2.42175273e-01,\n",
       "           1.36758542e-02, -3.78804297e-01,  1.12262454e-01,\n",
       "           2.18491104e-01,  2.30515151e-01,  5.61208289e-02,\n",
       "           1.06367322e-01, -2.05775586e-02, -3.67171708e-01],\n",
       "         [-4.23603434e-01, -2.57814570e-01, -5.69925794e-03,\n",
       "          -7.22925464e-02,  7.84545408e-02, -2.76215484e-01,\n",
       "           1.94344768e-01, -3.89348494e-01, -2.10042805e-01,\n",
       "           8.79007404e-02,  1.32646071e-01,  9.14340398e-03,\n",
       "           3.18588573e-01,  1.53023611e-02,  4.41578849e-02,\n",
       "           2.00642986e-01, -1.15938075e-01,  8.59744835e-02,\n",
       "          -4.77453937e-02,  9.32420403e-02,  7.72095945e-02,\n",
       "          -2.34277187e-01,  2.34552375e-01,  1.06016443e-01,\n",
       "          -3.96880033e-02,  1.87730180e-01,  1.43274338e-02]]),\n",
       "  'b1': array([[-0.12432595],\n",
       "         [-0.26057981],\n",
       "         [ 0.13652953],\n",
       "         [-0.00138851],\n",
       "         [-0.16887307],\n",
       "         [ 0.21799891],\n",
       "         [-0.22937835],\n",
       "         [-0.01121111],\n",
       "         [ 0.13465372],\n",
       "         [ 0.40340082],\n",
       "         [-0.36803927]]),\n",
       "  'W2': array([[ 0.69685509,  0.79988548, -0.6674788 ,  0.07622928,  0.79461253,\n",
       "          -0.78502573,  0.92243806,  0.14673577, -0.62059426, -1.16371674,\n",
       "           1.00399237]]),\n",
       "  'b2': array([[-1.00354299]])},\n",
       " 'learning_rate': 0.007,\n",
       " 'epoch': 12000}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# modelin ürettiği değerler.\n",
    "d1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ÖRNEK 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train boyutu:  (8, 614)\n",
      "Y_train boyutu:  (1, 614)\n",
      "X_test boyutu:  (8, 154)\n",
      "Y_test boyutu:  (1, 154)\n",
      "Epoch: 0, Cost: 0.693039\n",
      "Epoch: 100, Cost: 0.674404\n",
      "Epoch: 200, Cost: 0.662895\n",
      "Epoch: 300, Cost: 0.655513\n",
      "Epoch: 400, Cost: 0.650314\n",
      "Epoch: 500, Cost: 0.645890\n",
      "Epoch: 600, Cost: 0.641014\n",
      "Epoch: 700, Cost: 0.634470\n",
      "Epoch: 800, Cost: 0.625117\n",
      "Epoch: 900, Cost: 0.612249\n",
      "Epoch: 1000, Cost: 0.596025\n",
      "Epoch: 1100, Cost: 0.577584\n",
      "Epoch: 1200, Cost: 0.558640\n",
      "Epoch: 1300, Cost: 0.540825\n",
      "Epoch: 1400, Cost: 0.525199\n",
      "Epoch: 1500, Cost: 0.512151\n",
      "Epoch: 1600, Cost: 0.501594\n",
      "Epoch: 1700, Cost: 0.493204\n",
      "Epoch: 1800, Cost: 0.486601\n",
      "Epoch: 1900, Cost: 0.481428\n",
      "Epoch: 2000, Cost: 0.477385\n",
      "Epoch: 2100, Cost: 0.474228\n",
      "Epoch: 2200, Cost: 0.471763\n",
      "Epoch: 2300, Cost: 0.469839\n",
      "Epoch: 2400, Cost: 0.468335\n",
      "Epoch: 2500, Cost: 0.467158\n",
      "Epoch: 2600, Cost: 0.466234\n",
      "Epoch: 2700, Cost: 0.465505\n",
      "Epoch: 2800, Cost: 0.464928\n",
      "Epoch: 2900, Cost: 0.464468\n",
      "Train Accuracy: 77.36156351791531 %\n",
      "Test Accuracy: 74.02597402597402 %\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = pd.read_csv(\"datasets/diabetes.csv\")\n",
    "df2 = data.copy()\n",
    "\n",
    "X = df2.drop(['Outcome'], axis=1)\n",
    "Y = df2['Outcome']\n",
    "\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2, random_state=42)\n",
    "\n",
    "scaler = StandardScaler()\n",
    "scaler.fit(X_train)\n",
    "\n",
    "X_train_scaled = scaler.transform(X_train)\n",
    "X_test_scaled = scaler.transform(X_test)\n",
    "\n",
    "X_train_scaled = X_train_scaled.T\n",
    "X_test_scaled = X_test_scaled.T\n",
    "\n",
    "Y_train = Y_train.to_numpy().reshape(1,X_train_scaled.shape[1])\n",
    "Y_test = Y_test.to_numpy().reshape(1,X_test_scaled.shape[1])\n",
    "\n",
    "d2 = nn_model(X_train_scaled, Y_train, X_test_scaled, Y_test, \n",
    "                  hidden = 4, epoch = 3000, learning_rate = 0.01, print_cost=True)\n",
    "\n",
    "plt.plot(d2['costs'])\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Katman sayısının artması, her zaman daha iyi bir model meydana getirmeyebilir. Tamamen verinin yapısına bağlıdır. Yukarıda bulduğumuz sonuçlara göre, lojistik regresyon bu veri seti için biraz daha iyi sonuçlar vermişti."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ÖRNEK 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train boyutu:  (12288, 209)\n",
      "Y_train boyutu:  (1, 209)\n",
      "X_test boyutu:  (12288, 50)\n",
      "Y_test boyutu:  (1, 50)\n",
      "Epoch: 0, Cost: 0.690517\n",
      "Epoch: 100, Cost: 0.637691\n",
      "Epoch: 200, Cost: 0.602198\n",
      "Epoch: 300, Cost: 0.538520\n",
      "Epoch: 400, Cost: 0.460557\n",
      "Epoch: 500, Cost: 0.378396\n",
      "Epoch: 600, Cost: 0.383012\n",
      "Epoch: 700, Cost: 0.338503\n",
      "Epoch: 800, Cost: 0.294666\n",
      "Epoch: 900, Cost: 0.253259\n",
      "Train Accuracy: 93.30143540669856 %\n",
      "Test Accuracy: 80.0 %\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import h5py   \n",
    "def load_dataset():\n",
    "    train_dataset = h5py.File('datasets/train_catvnoncat.h5', \"r\")\n",
    "    train_set_x_orig = np.array(train_dataset[\"train_set_x\"][:]) \n",
    "    train_set_y_orig = np.array(train_dataset[\"train_set_y\"][:]) \n",
    "\n",
    "    test_dataset = h5py.File('datasets/test_catvnoncat.h5', \"r\")\n",
    "    test_set_x_orig = np.array(test_dataset[\"test_set_x\"][:]) \n",
    "    test_set_y_orig = np.array(test_dataset[\"test_set_y\"][:]) \n",
    "\n",
    "    classes = np.array(test_dataset[\"list_classes\"][:]) \n",
    "    \n",
    "    train_set_y_orig = train_set_y_orig.reshape((1, train_set_y_orig.shape[0]))\n",
    "    test_set_y_orig = test_set_y_orig.reshape((1, test_set_y_orig.shape[0]))\n",
    "    \n",
    "    return train_set_x_orig, train_set_y_orig, test_set_x_orig, test_set_y_orig, classes\n",
    "\n",
    "train_set_x_orig, train_set_y, test_set_x_orig, test_set_y, classes = load_dataset()\n",
    "\n",
    "m_train = train_set_x_orig.shape[0]\n",
    "m_test = test_set_x_orig.shape[0]\n",
    "num_px = train_set_x_orig.shape[1]\n",
    "\n",
    "train_set_x_flatten = train_set_x_orig.reshape(train_set_x_orig.shape[0], -1).T\n",
    "test_set_x_flatten = test_set_x_orig.reshape(test_set_x_orig.shape[0], -1).T\n",
    "\n",
    "train_set_x = train_set_x_flatten/255.\n",
    "test_set_x = test_set_x_flatten/255.\n",
    "\n",
    "d3 = nn_model(train_set_x, train_set_y, test_set_x, test_set_y,\n",
    "              hidden = 50, epoch = 1000, learning_rate = 0.01, print_cost=True)\n",
    "\n",
    "plt.plot(d2['costs'])\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Üçüncü örneğin sonuçlarına bakıldığında, önceki derste işlediğimiz lojistik regresyona göre, bu YSA'da 50 nöronlu bir gizli katman kullanılarak daha iyi sonuçlar elde edilmiş oldu. Lojistik regresyonla kurduğumuz model, aynı verilerle %74'lük bir test doğruluk oranına sahipti. 2 katmanlı bu yeni modelde ise %80'lik bir başarı yakaladık. Hiç şüphesiz, daha fazla veri ve daha iyi bir tuning işlemi ile daha iyi sonuçlar da elde edilebilir."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lojistik regresyon modelinde denediğimiz, veri seti dışındaki 5 resmin kedi olup-olmadığı konusunda, yeni modelimiz lojistik regresyon'dan daha iyi doğruluk oranı verdi. Fakat 5 resimden 3'ünü doğru tahmin etti. Daha farklı resimler iki model için denenirse, lojistik regresyon ile arasındaki başarı farkı ortaya çıkarılabilir. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelimiz bu resmin KEDİ OLDUĞUNU tahmin etti.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelimiz bu resmin KEDİ OLDUĞUNU tahmin etti.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelimiz bu resmin KEDİ OLDUĞUNU tahmin etti.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelimiz bu resmin KEDİ OLMADIĞINI tahmin etti.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD7CAYAAACscuKmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO29a3BcV3YeutY5/cL7QTwIECRBihTfT1FPShxKskayNBnZiSe2U07GzqSU8nVujStJxTPJrVT53rq35laqUk5VUkmpEtuT2J544rFGsjIZiaKk0ZsUJYoUXyBBEiRBggABEECj3+ecnR9onPVodAMSSYAzvb8qiauxd++z+/Tefdbaa61voTEGLCwsfvHhLPUELCwsFgd2s1tYVAnsZrewqBLYzW5hUSWwm93CokpgN7uFRZXgljY7Ij6DiH2I2I+I37ldk7KwsLj9wC/rZ0dEFwDOAsBTADAIAB8DwG8aY07dvulZWFjcLkRu4b0PAEC/MeYCAAAi/ncAeB4Aym52RPxSvywzvytFmIBEWPhwrkNjmKD8GPwV6nkwRSgAGsNRPYMFzgtLrjA3ahK14nXe8+lavifaDLs/Fa+NdO2IK5eB79P4hn3OCEpF0AhZfmZkfePRaCjX1tSIfsnpFLuunLvjxGh8/lCqsIzciLyn2Vx+zn4l3y2Wb+TX1s9G0ZUNgupeBQG/p2rNse+s0oqotKpmv09jAjDGzDnMrWz2FQBwhb0eBIAHF/pmvvkAQHwS3/iiKRGrpzaPvry8n1Wj0iCuutnNNTRGLpuhMdTmCPiaUqNH3XgoZ70czS8SFf0yHs1Lf7EOWxAuRlUb+zFhq2rjPTtEv6s3JkJ5Ojku2gqFNF2b/aghyk8TYfe/s7lDtE1M0/h5j+5VW02dvFZAPzReID9nLEYbtbe7J5R3b9ki+r31/uFQTk1nRFtt7YpQzuXZj5pTEP3QpWs3LouLtr6LtESRbThXGbCRKLs/6l4V2Ps8T/64umyVuGwdRKLyXqUydE89T/4A+QG9lleW95T/GBo1x1hs5vvM5/WeINzKZp/r16PkxwcRXwCAF27hOhYWFrcBt7LZBwFgJXvdAwDXdCdjzIsA8CLAjBo/+4MUBAtTNwHk03Ch7wqUvpXO0q9nFPh4sp/HnvSu+j3zy5gQaU8+kRY6LwPqKWHY1+HQU+JM/zk5D2AmidKCfL/A2uhaDsonng/05B1L5URbYKivz27PtaT8nAHwJ6ycB6TpdTJL2gb/HgAAhm6SZuIrkyQdjIWyi4lQdhylIudpzKkrUtMBNiY3SSCQSz+CdD8KgVL92RM1UKaGG6FxuPbseeoJW9EU4Fouu1ZJR/rcJZbMAizHWzmN/xgA1iPiGkSMAcBvAMArtzCehYXFHcSXfrIbYzxE/CcA8BoAuADwx8aYk7dtZhYWFrcVt6LGgzHmJwDwk9s0FwsLizuIW9rsXw6o/p19yV6XuI/IIOHuKu26quSK4yPmmZ0bBOWNH1+7SAJlly4I+hyTjymtKGSn/eAwG7VGnpYHObKdAz8t2hyHvA6Nje2hvGG7PNFfe+/6UG5oapBjsKNqv0A279joDdHv/OmzoXyx77Rom5oaDuUcG6N/cED083zu2pPLMZMn+7s+2sD6NYl+sWgjzXF6UrQZ5jFwHPoufCPt8hQzsbXnwmffu7Z7ucfDD+jsQ58FOcz7EVHem8Cj6xlD5yB6DG7bB/qMZAFeXBsua2FRJbCb3cKiSrCEarz8nXFZlJxvpAuGayhcVf8iEXS8Z4Gp7pWj3ZSZYOZWwbEkCq+8g5BHVmnPCjcpHJcFaNT1yI6YDMXahHSpbd++LpQ379kTylOmUfTrP/p5KN94Qx67cHdebQOZBd2rV4p+T//K10K5vunvibbD7x4K5UMHD4by9IT0znqGRdCp++by78nQ/ZjOy4AskyNTJqvcoB4ytZhpvg3RhOgHSGuuEMjxeYRh4MuAHh31F/5dr4GgfJScES5dNiW1R3ym4psFRl9y2Ce7hUWVwG52C4sqgd3sFhZVgi+d4vqlLoZoZt0aaKRdFGU2aiGQdlFtnJIKCgVymeQqJMLIEESAmMttNDLeCr50YRjWpjOXHD4m9xSqe8jdIjqc1WUuNR0yjOwIJVLTGcq1bdtEv2ZmRz/2+B7RlkzTXI6fuEp/Tyl3UorcaLmht+T8DU/8YO4k5RKtiZMLbGXvPaJt33NPh3Lv+rWh/NMf/Fj0+/j9AzQnZr8DAMTZmoi5lC2X93UCEftuTVK0ZQ2tEZ6pmHDkGAmWLZcqyDOjCHNFGuWqLbB7wjx7JaGu/LWrktI8Ni9kg2hXr8f2RaDWVTw+E+6bz2chCPw5DXr7ZLewqBLYzW5hUSVYAtfbLMqrOXMkyn7x4VVakIiK4mqUDuRjf4igNAW4Ws/H85U6LtVzPQZT1XUkFXPrONHWUF69ap3o98AT94fy2Wy9aOs/Ti61/CRFkxmUUXLNbd2hnE0uE23jLPotAFKDa+rkZ2msJ9W6/9wx0Xbhj06E8o77Hg3l3/rd3xb9Vq9ZE8qv/uBPRFvepxxwYPfNUdFjnBcgHpHkGHlm9vHvRZsknk+fTbtSPUHmIcFNOJ4VWRLhxtaVryL0XMMiFgM+nhyDv0tHG2o33VywT3YLiyqB3ewWFlWCJVDjTfH/So1ihAElWjz7g1S/vgAHnYhcq0CBIQgfpLqlX4dzUq85vZSvrsWtlVhMRrXlWdLGivbloby5t030e+/TIRp/5QbRVtfUS/MYokSS2g55ou9miRgiqNko5z89Tf2YGZLPSe9HCtkpsuKxMyzR5Ngnb4by2P91VfT71h/8fijHor8j2n74Jy+G8jRL+HEUnVeCqcyePohmhCAu8DUg106eRy+qZ2CBjV96ys6+X+Rqt14rvK38M9aw6FFtaqDwNukx5qd1sU92C4sqgd3sFhZVArvZLSyqBItus8+aPKW/MuU5nHl2m7ZjykG7T/iYHrOLTEmGHXUstdHnJqp0UGcn8Qwn9Uk58aAnP8uKFfeF8rP/+B+G8okzkjQimyF7u8mRLrV1O1aF8niWItwGzn0s+jXUTYXytOSbBJ9llbUtIwIMcCThw8QERbx53oRoc5hriLvDbtzoE/1e/H/+v1D+p3/4r0Tb8JXnQvmt118K5cBRXPkskjISlecgbkDX9g1lxBlFXhGwdaBJS4QDDHUmJG/l0XQqq5NvtRK2SJb1xvnlS9Y6I7RUZKURiMJ8sE92C4sqgd3sFhZVgiWMoFNRREzt80xB9WUuBxEJp0dk1TnUR+MVaDz2G6crx0jXipwjT66JMlVMK2VZpiK6iq8dGV97PCa51O7f+1gon+uj93Vv2SX65a4T99vWFS2i7eoFin5zk5+F8upOqWY3NJN6vn1Xr2jLFIivrsDcbfnpKdHv8EcfhbJfkEksHjD3HVORaxMyWu/mBLkRf/gf/pNo+41/8n+E8pXTRFx8/dpZ0a+xlu5pbY0slTUxyubMzTelSnMyCP0E5GWvPOXiQmehz0ver4IaX2GEL0LWMt8MLCwsfoFhN7uFRZXAbnYLiyrBktnsJeGy3J7SGXFcrki2we1oOT7n/pb11sqHHfLwVQCAqOAdpzEKJXzyPDtOZScxV9CWLbLord/QFcqdOygbbF2XtHOnE2QPH/7+y6KtkOoP5T0PUIhs91pZPfXs2cuhfPmiJIEssJDTtk4iu1y3RWbfbb5vZyh/9NY7ou2zo/Q64lKoa21UnjHwjMEzp4/IMd55N5S/9neeD+X/+u//SPQLCnSukFAVXnmIrM/cU6jWkcO+d1dlO4pAV00SKuoMlCeV9H3OKa8JU2Txa0J51+8XCRWfxbxPdkT8Y0QcQcQT7G+tiHgAEc8V/22pNIaFhcXSYyFq/J8CwDPqb98BgIPGmPUAcLD42sLC4i7GvGq8MeYdROxVf34eAPYX5e8DwNsA8AcLu+Ts78vCudYDVnaXu950IJIoB6U0IM4jlmPaeUkwE5Z9AQEjvSgwFb82KqPY8rzkr5Gut7ZW4l7vaJU87BnSzmE5K/G0tUMSMhxLkNrauU5yv63qICXLZ+rtjxT32+goua8aa+Ucsx6pwqc/p8/2aVSWoVqxkkyNr/zyftG2eTOp/G+9TNeujcREvwTjkyvUSPX52PuvhvIj//Jfh/L27TJL78IpKj21rlVmCA4xrZ67RNFRrsI8dYzKaUCSlYTWJcc4AkZy4bryOdq7agWNn5DRbjfHKUIyk6Z5FApyT2QL5TNDyZwor95/2QO6TmPMEABA8d+OefpbWFgsMe74AR0ivgAAL9zp61hYWFTGl93sw4jYZYwZQsQuABgp19EY8yIAvAgA4DoRUxufUQszeVl9NAh4aRvNAcbbFlZaySmJfiPwBJfAaOWGUfnqa7FB7l1Lp9TpadlveJSi1RwjVbZ7194byjcvj4q29dsogm51HZ3AH35DnpYjEH/cY8/eK9o+P/gKve+jo6GcScnoN56oUV/bLtpq2OfmrN65jBzjytn3QvlHV86Jtq//+q+G8t/+tW+E8rs//ivRrxCQOu2qo+6JMVpWn777QSjft2+f6Hf57BmS8zdFW559gIJPppHrqBNxlnSSVyWdOAV1Cb04o0SPsihNHZk5nWJmQ0auzUyGlZ7ymNcIKkFXeC1THZnPteJ45fEKAHyzKH8TAF6u0NfCwuIuwEJcbz8AgA8BYAMiDiLitwDgewDwFCKeA4Cniq8tLCzuYizkNP43yzQ9eZvnYmFhcQexqBF0Bgx4RZdVbY0kGUimxkVPjkCUtNXRanJ8eo9sy+QogskIfm99rfLEgJvXka0ceORCqmuQdlJ9huzESE7yunfUk2vsmpGxSLlJVtpqlGz9D//mXdFv/9/eG8pJFf12+GeUiZYPyJe3sku66DZufTiUWxtlhF6NS58nUUeZeUPXJImGe5ls5U/PnRFtP/qLH4by7/w2lXPu3bRV9Dvw0UAoZ0ASWhpmv3508O1Q/gff/n3Rz6uh8mCnrg6ItkSUziOirIRUIiaz4woFulcFX54nCSKUklJfZG9H2VlQTURurZoorZcAVfSoy2x9lkXn+4oUU5WlknOcHzY23sKiSmA3u4VFlWBx1XgTQDY/44KIRROiLREjVSybl5U4K6nuevxZBCUVWMsRYJQU9AmljpYu0bJ9OyWufH7sOo2Xl8kXOzb2hnJ743LRVttOySMNGzeJtpZVZCZcPEFurrq4jHArTJA6PXRBqnZBnko+rVlNlWA3bt0h+l27QAkzIzeuiLZ7mum7OJahJbLyHvlZHnz68VAeTUrV9+IpIpt4/a//Zyh//Rsy8vrUqU9DeSgrv4toE0UYXh+nxJ3Jq9dFv62bKeHnzZ+9KdryAbnb0CE1vqVJRtpNJek79DPys4BIcFFuOWb2GbauPE9+LxMT9L2oPBtRlsqNVOKqq5QkMz/sk93CokpgN7uFRZXAbnYLiyrBopNXzP66ZLIy66ixgUoUF5S94wec2Lxcor8MAdXlloVbjv+9Ag/9nj33idf9fUSOeGWAssY2bZDZa71rm0N55YpVom3YUPZTXUS65UYG6J5cOPJ6KK/pki6pc++Sm+jS9YuiraWZ7PuVy+lafUdkRHNNLdmv23ZtF233dJC7zQxQSO/gKenme+083Y+d+x4TbcM36Mzh+tAA/f3SsOi3aytde/yoXBPRWpq/yzj2+49/Jvpt3UbEHO+887ZoS/s0Js9YGxuXBJzpHL3OGkmkz3nkXWUr85pxHidFUaG/vAycr8JxxRLMc5KVSiQX8jltSzZbWFiEsJvdwqJKsPjln4r/8gwkAIBUhtxty1pkFlYqTW1p1k9npXFuubwqj8N55I0p77bgWU3NTVLNPnn0QijnAlL7ciijAf/nqwdD+b6t06Lt1AiZFz27zou2aaYWJ4eIkGF12xrRr/88lVDKgnQTbd5NavG1s3QP1twjzYn2dqaqT0p1cWCA7mszy+Ta/NXVot/QEcqqu3p4QLQ9+Shlpr3+v14L5ZELl0W/PfvIJXjijFTPp5HmH0GKsLx8UX7mh/Z9JZTr45JIxGPEE7VxcvfWxaSZx7pBTJmAOU6eUlLvYG4+eM1tWB8nd2YsLsdPZeniAYuaK/jSpRuwegraY+wUv6dK5Br2yW5hUSWwm93Cokqw6Gr8bPSaVjayOVJ3A0XN3MrU+qYmOrWfSk6KfpkUqfgOagICHkHHqJ7VYXxLC6nuUzfHRduTT1AE3ZFPaI6fHD8h+t2/cU8oG5Cca8DKE00PyfFHR34Uyg1MHR26LscYm6aT9VXd0hOQ9ijqz01QZFx3a7Po18JO41t7pLkCBfp2pq6SuTJ1Rp6Wb9pNyTRHXv9ItC2LkWnz0EoyIbLX5Il+tEARhZtWyCSZvghFALr3EO9c8uTfiH5BlkzC9ibJkDYxQmaDw828gqzimmf8cQUVsclPutFoNZ7Ti9PftZrNCVhcX61+lrUliFrUGBGHrwM5RrRYKTeL8jvisE92C4sqgd3sFhZVArvZLSyqBItqsyMgxNxihJcySHLcZlL21MgouaQa6sj2XLValiOqbyR3Ut9n0obMsJLCvMqur37veGnnEydPirZelvT1+D7K+Ion5GcZuTYWytG4dFetXU2EiCc//IFoW9VFNmoDUFbWyLCMOuPlhXtWSA710ZvUd90msl87mmRU2EP7aF6TU9IVFKTp8zQFRKKRUS5Gf5Rs5e3bJfFl/gZF3m3dsDaUBz78QPTzb9BZTW+DzO47ff5wKMc2Px3KKV/xrl+nLMAVnTJT8eQwEWFmDbns/Jx6zjk0pu/L9ZfgtrLKROOuLqyQnVngdnk6q1pZuWi2/mJxWS8gyDMXoCPvQVPzzHpJZ+U5Fod9sltYVAnsZrewqBIsMgcdQK6Y5KK54WXkj+L5Ytzfk0lS2YIrUv18cN8vhXJBuSZ4hU3O7dXSJFVTbl2MT0jX2I1hUitT06SKbd22S/SruZ/Ur5ExGcl3cYhUSRNIVSyYpr5dqyhqLpeWXOgbHyLSi1U9K0RbbGAglOsZz5rjSs61m6OkmrpKNU2wyrMO4xipV1x7cbZ8Yk1rRdvVT0mdjNVS9FhboyohFadrN/R0irb6swOhHI2TmaAjFoeuk5nX3CxJKaKs/FYN0PcSjcmlP5mj79Mo4hMwCyNPaXDpnqIjxzAui65TBIkRPhemxmsTM8lMXe0CzOdnTKpKiV32yW5hUSWwm93CokpgN7uFRZVgUW12FxEaYzM2VEzZNKNZspk0lzvP7o8z+2bDGmm3rO0i91LT0w+LttOsrG+BESHkFFnkjTGWzZaVbRGk12cvEFFiPiPdTrvvo2ufOCxdTZdGKdR1ZafM0Ir4ZNuu306ho/XdMgQSb5Jrz7s6KNrqHbJRs+PMfmtqEv3aOuj+t3bUiTY3YMSJabKPM+ek+64+Ti6q/IicYyZLn625lsZb81yP6Hf1DLnool2Svz56g7ICk6eowlh66pLoNzlJn61jucyYjLLn2XiOwqmjKN18HrPLC+oMo4atR0c9HwPkbcyFpmzqm2x9xxyVVccIJz32vtoauT7qaims2dfhskVeekGmqrCQ8k8rEfEtRDyNiCcR8dvFv7ci4gFEPFf8t2W+sSwsLJYOC1HjPQD4Z8aYTQDwEAD8HiJuBoDvAMBBY8x6ADhYfG1hYXGXYiG13oYAYKgoJxHxNACsAIDnAWB/sdv3AeBtAPiDSmMFxkC26D4oBIqHS1xTtrU2k9to725y3Zw6L11S/+3PiCRhy7a9oq11+SOhPH59IJRHhyRnOhRIvUNlTviGXGMpoMivs1dl6aP19xKBRE+HzEoby5KLaiojVeudO+izZbKk6l1WnGvtjNe9kJZjFPLMbcYINpq2SXdVZzNFIsZysqxTlJkrdet6ae45qfrGGZ99IS/diJ3LyXWIHs3Dn5LzGLtGEX913XL8vV3rQ/lzRjbRl5O88Zk0lbZKxGQ9Al5zgPMS1rjyOVcI6L5xVy8AALNqoDUmo9rSTP2vj9D4KbW+HTZIRF2bE8k7QHIkIu+pz8asq5GZitFi31tS48WcEHsBYBcAHAKAzuIPwewPQkf5d1pYWCw1FnxAh4j1APAjAPh9Y8xUpV8Q9b4XAOAFgMqUORYWFncWC3qyI2IUZjb6nxtj/rr452FE7Cq2dwHAyFzvNca8aIzZY4zZM1e7hYXF4mDeJzvOPML/CwCcNsb8W9b0CgB8EwC+V/z35TneXoJZbu28rqnMnvp1NdLuum8rMZ28e+RqKE9MyZpw/Lfr6Mc/Ey29q4lbfNd994fyVmXbD5ynGmjn+iQDzXSOru0bcqX4itzy+jCxsaxhdicAQDZPtvLITUmcOHaFXFudSNdKqNpgkQSx9Zi8tN1ihkJHXWQZUGl5dmCyNOcgLefhNFHYJ/PkQWRS2rKQZDZwm2TC6fklyoiboMQzSJ6cEv0aGaF6TVTaw/fv6g3lkU8/CeWGmCKVZKGuMVUq2WVrIsLCWWujkv0nWaB77xj5DOSh1uAol5ehe7CshubPWZMAAKJu+a0WsL1gGHVSPidJWcWeQXmvZokvS0oXMixEjd8LAH8fAD5HxNmTon8JM5v8h4j4LQC4DADfWMBYFhYWS4SFnMa/B+VLRj55e6djYWFxp7AE5Z+KhJPq58MT7jbZ+HkfuWcmprhqI/XbCMtWCowkCJicJF3y4GtEQtjeLjnZe9cTAeIzm7eItncOkKUyPEbc7dpVOD5FLsH8VL9oe/iRXwnlTw8rwswMRYY1t9BnyShVPZen+9MYk6r1dJLU+I5OiiZDT6qETpoi3uINMoIusZLMBIzTPKIZaa5ErrJSwzXKldpIY9atIHdbwpVOm3QTmWw5FblWYGWlO+ppjI6czPRzWTaY/i4CtpamCqRaJxVBSiRK/XJGfk7ugr2Wlep5TYK+m5OMoNRX82hs5pF92u1M88+y8k+tHTILcDrJohR9OUfPm3mfzXqzsLCwm93ColqwyOQVBvKqpA1vm0UQyN+gbfeS6reRVUj98KiMfuNJLY46NU2ww8uRFKlbhRF5Et0/QNFwW7fIZJpHHidyjJd+RNVTC0aOkfbIhBibHBBtK65QtNru7eqk/gapZtkUReg1NcqT7hvjdK8yWclPF41StGFdnFTM1lp5T2MRmnNsmUwecRvptDsYI3U3qkomxXrYPV6m+MrZab/LONzS1+UpMl6mRJ54XEYDjgzSZ5tkUXiJnLzf9c0b6IVaXwlm2iVqSC32Vb+Ux7wE6kjb4zUHNKd8nk7xs4q7jqPBo37aEcXHd1jNhDrFQZdh3HX5gkxKSkRnzKHbFkFnYWHx8wu72S0sqgR2s1tYVAmWoGSzN2+fTE7aRQc/ILfZqhXk0tm7R7om3jlE2VAtzZJgMZ2m69awCL1UWtqaAbPJ+voOibYt27eFclMdkTB4Zkz0SyTo2umUtA0/PfEWXbtTRhjv20qc506C7PTzF4dEv7Y2IrDHrCaBJMLFxjhlTTXWS3u7tptcWbF2aSuDYcSJ7Ptyu2RWmsmTrekqm90fYHzq18i+jCRk9FvTRorsi3XI5ThUT/NvbySCkM6rZ0W/1hrqh8puzhu6djpPRBmgyCU85m7zNXkKQ1yVczYBva+OZal5ql5hNkv3R2d8Zj1GJMls7suDkpgE2Zi+ct9Fix/but4sLCzsZrewqBYsuhq/ENQmZJLC6h5ivDrdT+6YZEqaBPfv7A7lqWmpPl+4RK6VRlZmaHKqvFmR8yQ5xsBFMif2f/XrNKejklyixiE13ovIzzI+RabG6Yujom1L032hvHEPJeucmLos+o1NUvTe5i3SfQejzFXGykPnQSYXAYtIc+rkMjBZcm1hEz0PnE5FDHGE7o+ZkBGLwRRLpmmie5zIT4t+hqmtCNKl1raWVPybHpknHaNXRb9mFv1mPKnGcw8sTziJoiKGYAkoqHjjXZ5/oqI7OT88L98cqGjASITWnPEVsYXP3XJ8HiD7MdKLwFfRhoViPYYKmTD2yW5hUSWwm93CokpgN7uFRZVgkW12ByLuTAggKpuGmxqeL42VxnoK+9z/ELnezl6ULq+6WmYnorSVu5fTR708yN+nf++4PSXdJx8fIrfZ1q3Ub/POB0S/ZlZWGjLy7GDqOmW6NSj3SRfzSiVYZtuGHRtEv+EblB3X0CGzyCIxOpuoaSa7NDclbeqrH9N5Qdd2eQ/iDWTPuw1kp2NMnm8Y5vIqDEl73p+kLDvnHrJJE72ScbzwDp13TF+QWYBNu3aE8gqgTLehlDzDiDnk1spre5u9jrBlFUW59AuGvqdS9xW527RFHLCFy88HXEeNz7LUjOKN52WaE0xuqpXZiNMZdqbhyLU5W7vOhstaWFjYzW5hUS1YZDXegCnybKOKROIc2bGoVAn7L5Fqsm0Tudeee+Z+0W9giH67UmlVuilBLp/GRopcS05J91fBI1XSgMws8gyNcfzzg6F87ozkqlvZvTmU16+SKngjy+zauLlLtLXFSDW78D5l9F3KT4h+U2lyP65aI8spNfcwl9rIQCjH22W0YXqIVF9vlVT9atvJTRSpJdlvVNGG7UypzUk++IBFJsZ62Hib5HwTx96mIfr7RFtqkD53wyaa47JuWZY5eY2yGBN1ch6AtJY8QWiiXXQ0vmvktoiyrL2aqHTZuYzLLsHaMoocw2M6vl776LAyWiyazq1T5Z/YHsnlpVlWUzNzjx2n/PPbPtktLKoEdrNbWFQJFl2N94MZ1dgPpOpYYBxpuZyMsnIcUq3fO0Rq95FjkrxiGSu11LNaqs+bthNtfXMbcazlFAnA5QsUJXf2+HHRNnyVToHzBVIx0wU5j4uDdCKuP0ue0TGfPyFv/4M7d4XyhgeoiuvNMxdEv8FjNMfjP5NVYp/61adCedVuiq4LFEmHV0vXLhSkyWMY7bHhSRs1UoXFHlJh88MqMi4gNT7I0P3wJ+SJvrOK1O7IRXnWHTB1N7a2N5Td47Lc1uh1iuS75xFprjTXrQ7l8SQl0OhkF4ednruOPI3nB9y1dTJpiI/jsxP3TF6uq8Y2RhCiDvsLjEMvylwGqBJmOLV0Ois5BbNFQg/Pl6f0HPbJbmFRJbCb3RapRZcAACAASURBVMKiSmA3u4VFlWAJs950LBIrgaOMGl8QAZAtNJ2SduL0Rcoou3JR2ttH3iNbq7Z+WSh3rlwl+q3fRlzxv/rNfyDastNkJx155+NQ7jtxTPTL5ehcIedJd1XGkJvopgzygw+O0+ecGKXoqV1fuUf0q2+kNw59fk60Xf2YItLqAzrDaFeuNydO9nCmIM9P8qxcNLKSRkFK2qHGZVzrfZ+KNmRnGnCUzhycM9LVmTxBBA0jJ6U7Kb+XIidv/vSvqd87B0W/qxlyr3Wrsks7tz0eylfeJxKQgiMzGqPsfEAFcELWZ6WsMtINyrPgOJlqriA/S31A6y+noip5xJ7LXHSTEzJCdDJJ19a2eSIRLRlLY94nOyImEPEwIh5DxJOI+IfFv7ci4gFEPFf8t2W+sSwsLJYOC1HjcwDwhDFmBwDsBIBnEPEhAPgOABw0xqwHgIPF1xYWFncpsFKye0lnxFoAeA8AfhcA/isA7DfGDBVLNr9tjNkwz/vNbPRQ6XXLqx8VRlTjM6ukRJ1hPGXsfUaVkEIWcVVb3y3aNm4lcokHHnsklHXVzw/fOBLKN4fGRVvMkJppUJaGKvjELZcwpPb1LJMRY4/to8jBloS8tjNBvPRclW6ISf64Vd1kvixbu020tW5aF8pN62ge+UvSJCmcIlfkxBsyijB3mdxcsQwzqdrk/R6/QBVvJ12Z+IEPPBbKfW/851AenJCcfMcK9N1GOx8XbQ8+//dC+X+89BehfOG65BdMxGntpD3ppuRJLJrLna+5rMfUeE+aPK2N9B06KkmG89j5olKr/G7TWTJbfaXGx4uJMKnsBPi+N2c2zELrs7vFCq4jAHDAGHMIADqNmakPXPy3o9IYFhYWS4sFbXZjjG+M2QkAPQDwACJune89s0DEFxDxCCIemb+3hYXFncIXcr0ZYyYA4G0AeAYAhovqOxT/HSnznheNMXuMMXvmarewsFgczOt6Q8R2ACgYYyYQsQYAfgkA/n8AeAUAvgkA3yv++3L5UcLRgH5ftE3NzQxtz7tl+qnSt5ysr+TarJYc40JHxWPPSz2nUtJ2O3qYfs9OHj0aypu2q5pwT5E9HzfSVs5eIRv1yllZY22MeYMmkmTzXb0p3UlvHPgwlNe0ynLOW9YTyUPnSgoVrauTPPo13fQ62iXtUD9Kr9MsvDV1WobtZnktthuS4zzdT+5IN0LnFliQWWnjN+hcIVkjXVLNTZtCeSJK7tIsSrfZNLNtLw9fFG3pN+le7XqYnjcXf3xU9mOuMk/XemM2dV7ZyhFGApljNf58xRufyTHu/IgcH8uQTAYqXNZhn9NTZaWD2etVOIJbiJ+9CwC+jzMnaw4A/NAY8yoifggAP0TEbwHAZQD4xgLGsrCwWCLMu9mNMccBYNccfx8DgCfvxKQsLCxuPxY/gq6oZqA6LjAV1HPJ1c3dZpX44+RH4zxfnD+8lLOLtZW4B8n1kc2R2+zzTyV32vnTRMKwa/te0fbsU8Sr9sjj0lN57TKNf/5z4ka/flW6mtJTLOMuJV1Z505RhNr4KGWb7Xjuq6JfypDa7Q9I0oggTpF3QZRFlrWKbjD5CUV4xZj5AAAwOkjfRTZPpow7JaMeJ/Jkonid0tWZniaTJxMn0oskSJMhhzR+VhGO9A2QOdGzme5BXLnQUjke8aZ47BxeDksTT/BsuUjZfrWMl9BV/HcZxi0XY2QYiNIUSLMSUqWr1g+lcrCx8RYWVQK72S0sqgSLrsbPqs2VI/e+3G8QP63UCQFBwKiCuaqjqnmKJBxHR9exBJGAeMM875roNzVNp9SHP1ZtV06F8q4d94q2jhW9ody9hlINVq+XhAkRj05i41HJU1ZTT+qicUlFblojT/6hQKpjZmxKNPnnKSTCS9CJeGy5TKZJdVCCSP6GLMk0miRzYnKKxtffe5Z9Z4nmdaItnSKVdjJFplJSeTiShr6XlCdJHRobyOtgkNZAwZOJKtwE9I1UnyXruZw/J6wQ7wG9ruh9nq/579i12boqKf+EPOlGjhELTRlLJW1hUfWwm93CokpgN7uFRZXgrinZrG2cSj1Lpbmw0Gy+Sr932mhyWAtzs6hoqYCVEvKDG6ItmTofyu8fksQT9Q3kXvKZuyqmzh/a68httLpntWi7d+e+UB4ZIhdVTbck6cgM05zbmmUOU3MNXTvJIr9iIKPwpjPkekuOS1KHQZZ9N81KFBdcdQ4SoftYk5XEnWOH3g/la2N0JnDDl2NMMZ73hCPPJvY89GAoXxo4HcqZvDynMM7cdjOAPMcxnnYL05rwxbmQnGNyiqIIXb2smCxLNsu1GbBIPqPPFWB+2Ce7hUWVwG52C4sqwaKr8bPlaXSQ/xch0WCjiVfcNeYovwU6zF3DIp0cpSpxwoBoTCaZ+AVSnQKWiGB86cYxHrm8HB0t5VEEWU7xmY1MU6RcPEaJH0FeupPOMw74qcxK0VZTQ+6q6SSpqh+9JJMSG4HmjNskeYVhRA5ZpsLyEl0AAJksqdbn3vhT0TbNuNSmgOaUUeqnYW605mkZKcg8jJAKyMWYcaXrLfBpvvew0lsAAC5LOvnsM0qKAUclo4j18gWegchFVkJKEVQ4rM1ROrfvkdkQ8PWozDe/Ar/cQvaPfbJbWFQJ7Ga3sKgS2M1uYVElWPxab8XwQlPiLAhEPwmnQhsbIeBuC/nRXO4+YSGynnZhsGk5qg4cGB4uy8YLyrtBDEhChukshX1GVE2xHLPdXCC3VmODzCjL5ch2vsDKMgMA5A79KJQ76onAsrtBklZO+zSPuiceFW1ugj7n+AARVsRWSzayccbbn6mXbjk/zmzsSTovSKG0t6cZZ30UZehvnhntHnN75lxJuB8wvva6RpnNdmOYzgGy7CwFXbk++HfmonKvIV3PVSHU3FUWjRJhZkStvwg7Z/DUGY9Y+cz2Lgn5ZmtV11bwiu5eY7PeLCws7Ga3sKgSLLrrjdSM28EbX+E6SgXipYfRJXUIUZUQ5px0qqx0wNQ5oUah+izsdTwmudB7OrpCeWL8smhraaZMtxuTFHGVK8hoL5ephK1tMlNsW3dvKJ88R1zuazpkpN30KJkJn736l6KtsZZU8qkB4nRLLJPXijRSFlx83Q7RlrtO0Xtx5sqbTMqIwlgNqcXZyDLRNjRNEXUppiJPqarEniFOuivXJU/e9vt20rUi9LkKRrozORylqrsO1RIIlFnGDQDHZbUJHOmm5BF1JQyLzE2XZ1FyjqNcy2w9Omr82TlXiqSzT3YLiyqB3ewWFlWCJUyE0Wp7pQgg3rfSyTw7IQep6/lC66ZT9hIOOhaFFyj13HFYooOIANRRYfQ+NypVQidWfvy6BJkJN6cZ91sgvQIplsQRj8nxN99HpaEGx+gk+lj/adFvZTdVhr02JDn0co0UoVfTSsk50RaZMNPUS9Fqo31vibZMnk7qx9J0Cp4FqX5ilEyXs0PDoi3FohQzPr0vo8ozeQGp5KM3JYlGzqPowLoGIgEZnZBceK5IclIltRzuhZHfWcBJL5hXQJOiOCxiD1GPMXdyV0Sp8R6KcD0BmwhjYWERwm52C4sqgd3sFhZVgsUnnCz+WynSpzL4+5TdL0o2lx+fJ/4bTTjJ7CmdORcgtzeZraxK8fDxA8X97TnM7ReVv7U8qYwTYEznpbuno53s6B0bZOmpd956J5T376RyR2cuSDt3mpEwLGtdLtocl/jJUyk6Hwh8ydde00HRatFOeXYwfZYy+jIpRvToSPLM0SS5GMd9WRI6cKhUlO/T/PPeqOjns3vV3NAl2hK1dFM9FpHnqmxH8Z2ptWn8HH8h2pD1NXw7GfmdFVh0ZCGQbYKwgsm61FRg5ia3BADwi2XDbwuNa7Fs81FEfLX4uhURDyDiueK/LfONYWFhsXT4Imr8twGAH+l+BwAOGmPWA8DB4msLC4u7FAtS4xGxBwCeA4D/FwD+afHPzwPA/qL8fZgp5fwH842lXWK3BqVuMRULdfknLBfBVJ7EABUBAeLcBBjG1+YEJTo0tksVOdFEqunUuPytraml8etj5Iabykn1bceOh0L55riMrtu9ldqOniayhs7lslwfZkiVzBZUFGHA1Oc83dNVKyW/W5K7ubqkydPaTKQafpZ9lkDeU6/ASjzlpTss8DnnGosoBNWPmWz3bt0o2iYnyZyYnibZUQwSPnOpGdTRl2wear1I/ji6p9qxLE1C5ZZjJgUyUxF1Qg6P6NSuveKYtyOC7o8A4F+A/AydxpghAIDivx1zvdHCwuLuwLybHRG/BgAjxphPvswFEPEFRDyCiEfm721hYXGnsBA1fi8AfB0RnwWABAA0IuKfAcAwInYZY4YQsQsARuZ6szHmRQB4EQAAdeiQhYXFomEh9dm/CwDfBQBAxP0A8M+NMb+FiP8GAL4JAN8r/vvyHZznAsHDCZVtxfjKubuNk1QCSG5uRBna6XLSSmaD+ZCQ/Vgmna/s+akpytACTbqZpbDPe3opS+2ebfeLflevUeZYW6fMZkuyb/TjQcoa213fK/pNs9LJ9RE5fx7GG2NcEP1910W/Gz7Z0T0ZOX7OJ576KUO28uWMPGOYYCSeOVU3LR4hl5dhpJVRlAQVwLIYb47J0F+I0veUM+Tai4I+jyG5lHiC7k8sIts4cQknhFQmNXjM9eaU2Ozs/IdNJKYINnjMd6FMjblKuJWgmu8BwFOIeA4Aniq+trCwuEvxhYJqjDFvw8ypOxhjxgDgyds/JQsLizuBu6b8UyWgSPzn6rlWTFhmkWrjEVM+07H0IYIuqyNH5649UitjsUbRL1cglW2KkUQAAGTj9Flaa6U66jI18OH9+0P5wgVZ9rmznaLEzl/sF22jPpE8tHR2h3Lf1T7Rb10XZaxNTEjVN81cSAlWCenK8EXRb1kNueJGLkleOEjTvUoxnjnPSJPBZ6WbAlBllwy5MB1GHIIqkixgqm9Ds/wuLl2lcluijJOcrYic1KXIuNssEZUcer7HSz6x6Ddlovl8PSqXmhPha5VkV9c+YHJEEWw4rlPaScHGxltYVAnsZrewqBLcRWo8/91RySkLeo/sqbUZHkHnRKg1UD0L7DQetBrFTAHOA5eISwrkQoFO3DWfWaKG1Oy6ZnXaHyHVN5kiFbaju0f0S6dpzJsTMsGlgangG1atD+W+/pOi39nLR0N5y0pJET05TqQXqSSdiDdG5XyDAp3oTyWl59UE9DkN40trrpX3Klmg96GnEoqY+eYZlkgCsl9NDSuVpb6z60PkkeDquV453EzwtZnA1oTnqSQWTinO1Hie+DLTxlaxSsLJFRi1NIsGRFfeb96mI/kseYWFhUUIu9ktLKoEdrNbWFQJFt1md4u2s1/i4qrkUuOuCllcicNR0XAcPrO7eAkfz9c2GJuXL4keuZ1uGO+4p4OZ+GdTpIHZHNlnWx/YL9q8CYpQe+PNt0P5+V/7bdFvbIrIG9yoLCudmiYb/q0jb4TyzvWyLPOly+SSunJdklGuXkauvTM3iYe9vU7yunMiyUJKkkXGY3TOkPUoci3jS9t+Kn8mlAMjzx8CQ67JgNnpeu0YlsGWTksCjILH+eFpHWkiCJmTpu1hZkerAyROOMndYYGaY56VfNKRmfyYwWfEFlmVOycy5zSxii3ZbGFhMQu72S0sqgSLrsYHRqf1z6ICt1xZx4Imr2AquXJvgFDx6X2c20yPqKfKcw98nrxQwoXHOd+luuhlyTRIZaWqt2H7I6F8bYw+y8E33xX9RoYpAeUrX/lbou3GNSKUODd4PJSP9MsM4w09u0M5k5Kqb/9NMhPSLElmkKntAACJCCvJVJDllOpidCfH8+T+SqZlxF+OVZM1SjWVmim98JQKXmARi/q74MQQ3M1qlNrLSUscpWbzNev5OnqPxsmyqr8FZQJyNVtHALosSYur53qO2iQUKFdVjb+9fJOFhcUvEuxmt7CoEtjNbmFRJViCks2zqBTgt1BCG00dWZ7zXRBPcOhab8JloufB+9LvpONqMgXq19wmQ10dRpZ48PXXRdvFc+dC+evP/1Yov/fGz0S/60NEAPHG66+Jth07HgjlezeSfOXSCdFvPMtIJJSbaIIRbLQ3ELWgX5AhoNPMjVjDyiEDACRTp0J5MkV2uh8otxk/31BhsNK25SHOugYfy5xT45ccvMy+p4SYpPxWMOy79krOiei1JJSQ60oTYnDwkFtObBGtMCftapsd47bwxltYWPx8w252C4sqwRJkvc3+vlRSkSu53srzb4sxFQlYJMKi35japEkuhKGh3HfIXR8sSwqM7NfWQZzpTzz3d0RbLEHq4wfvHBRtw4NnQ/knr/xNKK/uWSf6dbaTuni+/5JoO3b8UCi3tjSH8ratT4t+kxPEYxeJyWUwdpMIN66OUcmnVW2S725ymlx0TiC55dI5csXFYjSPeEKSV2QLLLouq/jgA15am5loIIkymprbaLysdCN6bAwe7aZLdQuVWUVi5n3OhafLRvF1XME05a49lQkpeOm5WaDctlHmotOrFp3bxxtvYWHxcw672S0sqgRLoMYvpPzTQlLxK6MkuZ8lSwRBuSg+qKhucdUvCPJMlsk0O3fvC+UPDvwv0ZbNUhTaV77+m6Lt9asDoXziMyKX2LD+PtHv4QdJrU+l3hFt/Rc+DuXOduKgO3v+sOjXuYzGOHvpuGjbvJqud32EEmEGx86Lfs11VJF1PCkrqzoJ8lDUM9lTpA5Nta2h3NYhK7Dm2Wl/Nk2RjtoU6Okhj0ffqWOiLQD+3TD6b1VlNeezSDujT/sZaYk6IeeJVDx6L6JMgQLzCrh6fXNSCn5t7VFyaR4RmNvEzGnvEoN9sltYVAnsZrewqBLYzW5hUSW4iwgnyxNPfBkbXmc/cRKDgJdkUhzevCRToFwfwOwzz6Ozh5oaWco4wTLFOtpXiLbDHxGhhJeSWWSNjIxxEiiz7dB7h0S/tmWdofzQg4+ItpuT5CpLs/OBtWtkCalshuzhgi9t1DOXKdpufTez7ZNXRL+xFN3T+gZZxDftU9vYKBFbBIwLHgDAZ2cf8XidaFu2jMpdd3XTfRRZYgBw/hwRYKRS46It4XACDH5eJJ9zcZbBFyivcJ6tnVROuvYcccZD28nz/bL9SsDLhFeIgcsVWElodZ4Ui8R09xIstD77AAAkYeZ0zTPG7EHEVgD4SwDoBYABAPi7xpib5cawsLBYWnwRNf5xY8xOY8ye4uvvAMBBY8x6ADhYfG1hYXGXAksS5OfqNPNk32OMGWV/6wOA/axk89vGmA3zjGPKq+SkZJTGxZWbo3ah8eg3peZEKVHDcDICT5Iu8DbOJQ4AYFgyQ9fy3lDuXiH53YbHqGrp0899Q7SlMqRaDw0PibbBi8TtPjhALq8GVYE14tC82tuWi7bOLlJ3/Txda3hEus0yKVKf21qkqdF/8fNQdgK6P8tYRB4AwMWr5B5sbuwWbekMRcPl8lz1lS4vL2DcbMqUc5GiHrnLC1V0pMdMg4jiWo9HyDSayrKoQe1WZc+90tXGo9oULyFbZ1yVzlZYV27JtXl1Vs5D76t+TFbrOxGfuXY2lwE/8OfcZAt9shsAeB0RP0HEF4p/6zTGDAEAFP/tKPtuCwuLJcdCD+j2GmOuIWIHABxAxDPzvqOI4o/DC/N2tLCwuKNY0JPdGHOt+O8IALwEAA8AwHBRfYfivyNl3vuiMWYPs/UtLCyWAPM+2RGxDgAcY0yyKH8VAP5vAHgFAL4JAN8r/vvywi45a/+UL0c7z4wWMHZpNpuIQhS2oXL5IecWl2cCyzspm23vQ8+GclOLtGAmmXfpgw8lWaQbkC07OHJdtP3qr/2jUL7YT8rTqaMynJXXVYvGJXFGvIZ45C+ep2sv75QZa5evUIZda5MMU922gX6Xjx5/K5S5Kw8AYN3KnaF8ffyqaCt4FN6KMHdIKQBAjJ2LBOqrdTjPO7OVHZWNKF4pe74gMtZYN2WY+6yWnCbHiLJadXFHnuOI+n/MFo+5cmvlWJiw5pSPsPOICCvf7CsfoCOyOuX8Z8/eKp3ALUSN7wSAl4px4REA+AtjzE8R8WMA+CEifgsALgPANyqMYWFhscSYd7MbYy4AwI45/j4GAE/eiUlZWFjcfixBBN1skr3m6OI8X5pHbGEEAXxMXd7HL6PO6X68NRaVKv6eXY+G8vsfvhrKacX//vQzlM3W0ybLM73201dCuXn5etH26YcHQrmNRd49+dRzot/bb7wUyjdG5Fnpxo3bQ7mQ2xzK2eyk6LfuHurX139UtG1mpaLq6mj+g8PSnOhup+i6fEG61HzOx4ZlsrpAOk81XRzPVOTuV2PKrwHtjBWcdOwCfgl/IbuuLhnOOd9LTAgWQcdUcEfNUQTQaTp41hiNkMlQ0AGcPpkCqCI/neLrShF4NjbewqJKYDe7hUWVwG52C4sqwZLZ7CWuN2YLoTLeZLhseZukvDWoXy9svIYGGR7a1ECsKqk0hcSmUpJ95dSpj0K5o02GorqM+FJnQvWdJJaZM+aDUH7+V35X9OPki7mczPL65FMisdzA7PLklCyp3Nm+MZRbmmXI7YULFLbbu2oTm7u0y6/fID54bctyuEihyrobD5fV5ycOUvYgt1EjKGsA8Bp/nrqnKMhA+RmAnEiUudRcNYZgtdHZa+y7SGeIdFOfTXC+fEetfV4/TlxL2f18BTv67GMBrjf7ZLewqBLYzW5hUSVYUNbbbbsY1z9LMHdJ5RlUIIhc8LXnHr/08zN+ckdmUD21/9dDub2dSA5P9X0i+q3uJZfXe+//VLQ9+MjXQrmlVUauHXj9B6E8OkIRae2d94p+W7dQhJuvbs27H70fyoXkQChv2bxP9Bsfo6ysri5Zour6CGWHTYyTqr5pnUxqvHCZzI7JKRkt7TBVOxEjco+IikDL5CmKsKDKZ/OMtZhDKn3EkVGDhhF+6nLOXBNOF4gPX0fyNcXIRNMlqnI+RQ7GI9LydRxaL9N5lumm1fiypcpLOezDsVGPQfPS6zYRn7nfuXwegkDHIhbHKzsDCwuLXyjYzW5hUSVYdDWeTm0V7zV7XRrVFoiehPIlpCKu5BaPstei2qsr57Gud0sopxXf2OWrRCix90Eqp/TIgzJq+GcfEkd7VEXhuQk6mZ6YkCxe27fuCuVXfvznoTw1LUsr9a6myLWe3q2irbt7TSj/5JU/DmXflyfYex/45VBOpeTn5Lz6EzeJd+4SU9sBALI5Uvd9xWPnMLOJm1BG8bUb9l07ak34LJKSV12NOg2iX5S9r2DkZykAP+1n60ip1fyEPFAqPl+PCbc8V2LW514ZnejFeOaUks0TaBwWNegpDkTOUa89OYn4zPrO5rLgWzXewqK6YTe7hUWVwG52C4sqweJH0IXcFV/WvfblzhjamltCeYrZqOvXSrLIpkbq18r42QEAOjvIRXX4yJuhfO78WdEvnaMMu689++ui7e33qDYbJ5AAAGhtIu7yxiaysSeTkmv9/HnKUhsZGxNtba3k5qpvpAjAoeuyFtvhT2j+LU0toq2paVkoc1u5oU72m0pdYf00bznPQOSEINJmB+ZOchwdGUeyx/jlHdCkj+U50/mq4hz1rgrlK4h5lV+LBUUoIV7xTDSdmSdsbE2wQfMyfvnoTv6uqCLHcOb23sk+83exsLD4RYDd7BYWVYJFV+PDyCXNoVXxXeV+k8pHv3GVDQCgq4NU0JYcqefZ7LTo98kJ4lzrZtzwAAD7H/1boZxipZuOnZLlkDkyaUkasXs7JadMTVwSbVGXVNrxMVKRAxVZ5nukxnqedMu9e4jU8/W95KLbfK8kG3r3PSLKuHlzQLQ9cN9XQnkiTep/PC7VbP69mBLmCe5uY6QLJZx/XN3X7lhanlxN1XUE+OtAkTqg8EJxTnap93K3n3a9uaxegFfiqubuQX6l8mtTX1tH85UDH1GX7IpG504w47BPdguLKoHd7BYWVQK72S0sqgR3UdZbpfeRzSTn65d2DiF/x9Z03RPKHS3E/36s/yPRL5uX9jHHg7spLJZHJH5y7APRL1FTx2RJOPnLTxMZJXdxAQB8+NFroeyy7LDauibR7513fxLKdQ3SPRiJ09lEbZzCLVeyMFoAgO7la0P5o8MHRFtzI425noXjfnhIZvCNTdKZg4syPDni0P3P+zyEVdvb9B1qrn9Zu4+RfqhecZey4Dwj3XK5gK5tWJt21yFbV9pm5+cKpfvFm6tbSdnnShWbBaFl+W7izsVViebZWm/TmTR4/q3VerOwsPg5h93sFhZVgiXgoPviiPBMIE4DXvFdsjWbYzxfjJC7qa5V9mMEBJzDGwCgqb4xlD85ThlgO7Y9IPrt2PZwKJ86IznZf/LaX4Zy+7I20XbxwolQfpSRXKxZJ6P84nEyDSYmZEmmwWvXQnnkel8oj4/K8tCFPKm0a9esE23NDTSvjz97O5SnktLNx0sql5RkYtlhUWRqti+jAXkWXIkLCrlrj5dnUqWM2dt0hJ4pExmnMyt5VlpUlVQuGO6yU649ns0mlHA5R67964y1hTEiSghefgDwfKd4nVvkjUfEZkT8K0Q8g4inEfFhRGxFxAOIeK74b8v8I1lYWCwVFqrG/zsA+KkxZiPMlII6DQDfAYCDxpj1AHCw+NrCwuIuxbyn8YjYCADHAGCtYZ0RsQ8A9htjhoolm982xmwoN07xPV/qNJ4n9/sBV6MqDSdVpaY6qrTa1kjUya2Nki56IkWJJbX1daLNZ/dqeJS403Zvl2r81eHBUN778FOi7fxFSn45+OZLoi3CzIbWZVR1dTotT5iffJxU/KiKSGtvp895+Rrxx13qvyD6ZZkaX9cgP+fg4MVQXte7O5QvDkhTIJcl3rlaRsoBANDeSgQTN6coSnE8KcdI53iCjo5qIyuTU1Xr6rquw701MnLSF6fzfL0oKml2Oq+TSvIBNzW0B4g6u2xMv4Knva3auAAABuFJREFUqISNmsnBl/SO1SRmTKps7tY46NYCwA0A+BNEPIqI/7lYurnTGDMEAFD8t6PSIBYWFkuLhWz2CADsBoD/aIzZBQAp+AIqOyK+gIhHEPHIl5yjhYXFbcBCNvsgAAwaYw4VX/8VzGz+4aL6DsV/R+Z6szHmRWPMHmPMnrnaLSwsFgcLqc9+HRGvIOIGY0wfzNRkP1X875sA8L3ivy/fqUmWZFQtCNJs4S61geHz7O/S+ljZSbZy1s+ItvMDp0P5vh33h/Lps5+Jfpevkq2cTMqst8cefiKUdRZZnvkVb4zdYC3SLj99+tNQbm2S7rtsnlxxqTTNf1WvdK8d/NnfhPK+NZIws5ClKMJclsZLp66Lfr6h8Z2cdAXdnKS2sSmyywvK9VYpZExmtwVlZPm6dK3w51n50t/8HEC7xioTq9CYUZfOXBwjvzNPZP6pGfLSZ6zVLyFe5e9RoyzA1F+on/3/BIA/R8QYAFwAgN+Bmbv4Q0T8FgBcBoBvLHAsCwuLJcCCNrsx5jMAmEsNf3KOv1lYWNyF+LmIoJPeiC+ZuMPexnnRr44Nim6jU6Sq5gtS5WysIzddSx1F052cklFsAKTORVDe4pEbFOGWqG0UbV999NlQvnBxIJTP9J0S/fr76TUvBQUA0Nd/mvU7Hsr7Hn1a9IvHaI7950+Ktgd3PxrKRz6hCECtftayJJ9IRKqtkRj1jsWoXwzkZ55KkStOu7Vkkgxzw6mjpkBEuEk4OPcSD4yKwhNj6k9a3tZAkazDrqsTfnRmTJnxXRZ56HtSjedmX01ClsCKFb/Pwk1Zi4DDxsZbWFQJ7Ga3sKgS2M1uYVElWFSbHREhGo0W5ZLWMjJAxKVQRtfhYZPSpil4FNao7bqaGAuH9MguCpRXxTBbLhqTtycSJTs3xsJD9z28X/T74NP3Qnl1rySNGGChqJs3SRLICCv/+5W9ZDcn4vKzfH7iUCgjyPDQndsoQy6VGg7lQkESa+575LFQXtYi3XcH3iJyjOkkc69FpN3peXTz8sq+TGbpdTxC42vPWDxO34sBxSnP7N4IsH4lWWPc9SZH4CQgjgi7liHI/I0uqvp8bMzS0sv0mpdeRsUbz8kmdIZgjLVF2TpNpVUdwvVEwLJ1q6zxl8vNrIM333wTysE+2S0sqgR2s1tYVAkWm4PuBgBcAoA2ABidp/tiwM5Dws5D4m6Yxxedw2pjTPtcDYu62cOLIh65G2Ll7TzsPO72edzOOVg13sKiSmA3u4VFlWCpNvuLS3RdDTsPCTsPibthHrdtDktis1tYWCw+rBpvYVElWNTNjojPIGIfIvYj4qKx0SLiHyPiCCKeYH9bdCpsRFyJiG8V6bhPIuK3l2IuiJhAxMOIeKw4jz9cinmw+bhFfsNXl2oeiDiAiJ8j4mezFGpLNI87Rtu+aJsdEV0A+A8A8MsAsBkAfhMRNy/S5f8UAJ5Rf1sKKmwPAP6ZMWYTADwEAL9XvAeLPZccADxhjNkBADsB4BlEfGgJ5jGLb8MMPfkslmoejxtjdjJX11LM487RthtjFuU/AHgYAF5jr78LAN9dxOv3AsAJ9roPALqKchcA9C3WXNgcXgaAp5ZyLgBQCwCfAsCDSzEPAOgpLuAnAODVpfpuAGAAANrU3xZ1HgDQCAAXoXiWdrvnsZhq/AoAuMJeDxb/tlRYUipsROwFgF0AcGgp5lJUnT+DGaLQA2aGUHQp7skfAcC/AEn0thTzMADwOiJ+gogvLNE87iht+2Ju9rnoPqrSFYCI9QDwIwD4fWPM1Hz97wSMMb4xZifMPFkfQMSt873ndgMRvwYAI8aYTxb72nNgrzFmN8yYmb+HiPuWYA63RNs+HxZzsw8CwEr2ugcArpXpuxhYEBX27QYiRmFmo/+5Meavl3IuAADGmAkAeBtmzjQWex57AeDriDgAAP8dAJ5AxD9bgnmAMeZa8d8RAHgJAB5YgnncEm37fFjMzf4xAKxHxDVFltrfAIBXFvH6Gq/ADAU2wB2mwp4FziQ8/xcAOG2M+bdLNRdEbEfE5qJcAwC/BABnFnsexpjvGmN6jDG9MLMe3jTG/NZizwMR6xCxYVYGgK8CwInFnocx5joAXEHE2TJqs7Ttt2ced/rgQx00PAsAZwHgPAD8q0W87g8AYAgACjDz6/ktAFgGMwdD54r/ti7CPB6FGdPlOAB8Vvzv2cWeCwBsB4CjxXmcAIB/Xfz7ot8TNqf9QAd0i30/1sJMPcNjAHBydm0u0RrZCQBHit/NjwGg5XbNw0bQWVhUCWwEnYVFlcBudguLKoHd7BYWVQK72S0sqgR2s1tYVAnsZrewqBLYzW5hUSWwm93CokrwvwGnEZfIj1wl9gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelimiz bu resmin KEDİ OLDUĞUNU tahmin etti.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# veri setinde olmayan resimlerle, modelin denenmesi.\n",
    "from PIL import Image\n",
    "from matplotlib.pyplot import imshow\n",
    "\n",
    "for kedi in range(5):\n",
    "    # resim dosyasının dizini\n",
    "    fname = \"images/kedi\" + str(kedi) +\".jpg\" \n",
    "\n",
    "    # resim dosyası açılır, üstte belirlenen num_px değerine resize olur.\n",
    "    img = Image.open(fname)\n",
    "    im_resized = img.resize(size=(num_px, num_px))\n",
    "    \n",
    "    # resmin piksellerinin standardizasyonu\n",
    "    arr_img = np.asarray(im_resized)\n",
    "    arr_img = arr_img/255.\n",
    "    \n",
    "    # resmin piksellerinin predict() fonksiyonunda kullanılabilir hale getirilmesi.\n",
    "    my_image = arr_img.reshape((1, num_px*num_px*3)).T\n",
    "    \n",
    "    # tahmin işlemi.\n",
    "    my_predicted_image = predict(d3[\"parameters\"], my_image)\n",
    "\n",
    "    # tahmin etiketinin belirlenmesi.\n",
    "    if np.squeeze(my_predicted_image):\n",
    "        pred = \"KEDİ OLDUĞUNU\"\n",
    "    else:\n",
    "        pred = \"KEDİ OLMADIĞINI\"\n",
    "    \n",
    "    # sonuç.\n",
    "    print(\"Modelimiz bu resmin\", pred ,\"tahmin etti.\")\n",
    "\n",
    "    # tahmin edilen resmin ekranda gösterilmesi.\n",
    "    plt.figure()\n",
    "    plt.imshow(np.asarray(im_resized))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bu modelimiz de, lojistik regresyonda olduğu gibi 5 resmin 3'ünü doğru tahmin etti."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4 style=\"text-align:right\">Zeki ÇIPLAK</h4>"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}