Relationship between Suicide Rate and Happiness Report Factors/indicators

Project Steps:

• Define the purpose/ problem
• Search for the data
• Download the data
• Explore the data
• Clean & prepare the data for analysis
• Analyze & interpret the data
• Visualize the data

Project Purpose: Identify the relationship between a crude suicide rate & the happiness measures, which are the next seven factors:happiness score, economic production (GPD per capita), social support, life expectancy, freedom, absence of corruption, and generosity of a country. These seven factors could be found in the World Happiness Report.

Data Source:

GDP per Capita and Suicide rates https://www.kaggle.com/harshav05/gdp-per-capita-and-suicide-rates World Happiness Report https://www.kaggle.com/unsdsn/world-happiness?select=2016.csv

Suicide Rate Dataset

Importing & Exploring:

• import suicide rates dataset
• explore the dataset

o see the first 5 rows (head())

o see the number of columns & rows (shape)

o see details about the columns (info())

o see if there is missing value(isnull().sum())

Data Cleaning:

• remove the unnecessary columns
• make the country name columns the index so we can join the datasets after
• split this dataset to two datasets, one for 2015, and the other for 2016

Descriptive Analysis:

• apply descriptive analysis on suicide rates for 2015 & 2016 to haver bigger idea about the data
• apply boxplot to the dataset to explore the min, max, median, & outliers for each of the datasets
• explore what are the top 5 countries for each year

The Results of the descriptive analysis:

2015:

• the boxplot shows that the top 5 countries are outliers, which means that the top 5 countries have suicide rate that differ so much from the rest of the world
• interesting that 4 of the top 5 countries are in Europe (Kazakhstan locates in Eurasia)

2016:

• there was a slight decline in the number of outliers (from 5 to 4), in the max rates (in 2015 the max rate was almost 35 & in 2016 the max rate was almost 32)
• Suriname took the 5th place, and as a result, there were 3 countries from Europe, and 2 from South America

World Happiness Report Dataset

Importing & Exploring:

• import world happiness report 2015 dataset
• explore the dataset

o see the first 5 rows (head())

o see the number of columns & rows (shape)

o see details about the columns (info())

o see if there is missing value(isnull().sum())

*This process run twice, first to import & explore world happiness report dataset for 2015 and the second to import & explore world happiness report dataset for 2016

Data Cleaning:

• remove the unnecessary columns
• make the country name columns the index so we can join the datasets after

*This process run twice, first to clean happiness report dataset for 2015 and the second to clean world happiness report dataset for 2016

• merge the world happiness report for 2015 dataset with suicide rate for 2015 dataset in one dataset called “suiciderate_happiness_report_2015”
• merge the world happiness report for 2016 dataset with suicide rate for 2016 dataset in one dataset called “suiciderate_happiness_report_2016”
• there are two datasets prepared for analysis, suicide rates & happiness report for 2015 dataset & suicide rates & happiness report for 2016 dataset

Data Analysis & Visualization:

• see the correlation between the variables in the suicide rate & happiness report for 2015 dataset using corr() function
• create a correlation matrix using heat map
• create a function that will return a scatter plot with a regression line, the Pearson Correlation Coefficient, the p-value & returns if there is a correlation, if the correlation is negative or positive, if it is weak, moderate, strong, or very strong and if it is significant or not

The function:

def correlation(x, y, t):

sns.regplot(x, y)

plt.title('Correlation between suicide rate &' + ' ' + t)

pearson_coef, p_value = stats.pearsonr(x, y)

print('The Pearson Correlation Coefficient is', pearson_coef, 'with a P-value of P =', p_value)

if 0 <= pearson_coef <= 0.19:

    print('There is NO correlation')
    
elif 0.20 <= pearson_coef <= 0.40:

    print('Weak positive correlation')
    
elif 0.40 <= pearson_coef <= 0.59:

    print('Moderate positive correlation')
    
elif 0.60 <= pearson_coef <= 0.79:

    print('Strong positive correlation')
    
elif 0.80 <= pearson_coef <= 1:

    print('Very stong positive correlation')
    
elif -0.19 <= pearson_coef <= -0.01:

    print('There is NO correlation')
    
elif -0.39 <= pearson_coef <= -0.20:

    print('Weak negative correlation')
    
elif -0.59 <= pearson_coef <= -0.40:

    print('Moderate negative correlation')
    
elif -0.79 <= pearson_coef <= -0.60:

    print('Strong negative correlation')
    
elif -0.80 >= pearson_coef >= -1:

    print('Very stong negative correlation')
    
if p_value <= 0.05 and 0.20 <= pearson_coef <= 1:

    print('This positive correlation is significant')
    
elif p_value <= 0.05 and -0.20 >= pearson_coef >= -1:

    print('This negative correlation is significant')
    
elif p_value > 0.05 and 0.20 <= pearson_coef <= 1:

    print('This positive correlation is NOT significant')
    
elif p_value > 0.05 and -0.20 >= pearson_coef >= -1:

    print('This negative correlation is NOT significant') 

• create y2015 variable that presents suicide rate for 2015
• find the correlation between suicide rate for 2015 & each indicator of happiness report for 2015 (7 factors/indicators) correlation between suicide rate & Happiness Score 2015 using the correlation function we’ve created
• see the correlation between the variables in the suicide rate & happiness report for 2015 dataset using corr() function
• create a correlation matrix using heat map
• create y2016 variable that presents suicide rate for 2016
• find the correlation between suicide rate for 2015 & each indicator of happiness report for 2016 (7 factors/indicators) correlation between suicide rate & Happiness Score 2016 using the correlation function

Results:

2015:

• There are weak positive correlations between suicide rate (dependent variable/ target) and Happiness Score, Economy (GDP per Capita), Family and Health (Life Expectancy) for 2015 (independent variables/ predictors) and these correlations are significant

2016:

• as in the previous analysis, there are weak positive correlations between suicide rate (dependent variable/ target) and Happiness Score, Economy (GDP per Capita), Family and Health (Life Expectancy) for 2016 (independent variables/ predictors) and these correlations are significant

Main Result:

• according to these analyses, we can state that a country that has a higher happiness score, higher GDP, higher family (social support), or higher Health (Life Expectancy), tends to have a higher suicide rate.
• As a result, we can state that the developed countries tend to have higher suicide rates