Airbnb_Listings_Texas_Data_analysis.qmd

---
title: "Airbnb Listings in Texas Data Analysis"
author: "Niharika Patil, Pratiksha Gadhe,Yashi Agarwal"
format: 
  html:
    self-contained: true
---

#### Research Questions

#### What are the factors and features of a listing that make an Airbnb listing more expensive?

##### SUBQS: 1)How do ratings and reviews impact the pricing of Airbnb listings?

##### 2)Is the location or neighborhood a significant factor in determining the price of an Airbnb listing?

##### 3)Are specific property types, such as rental units, more expensive than others?

##### 4)Do the number of bedrooms and beds affect the price of an Airbnb listing?

##### 5)Are certain types of baths (private, shared, etc.) associated with higher or lower prices?

##### 6)Does the room type (e.g., private room, entire home) influence the pricing of Airbnb listings?

```{r,warning=FALSE}
library(tidyverse)
library(dplyr)
library(ggplot2)
airbnb <-read_csv("listings.csv")
head(airbnb)
str(airbnb)
```

```{r}
airbnb_2 <- subset(airbnb,select= -c(neighbourhood_group,license))
str(airbnb_2)
```
### Data Cleaning

```{r}
# Define a function to insert ★ after the first period (.) in a string
# For data analysis, we need tidy data set. Each listing doesn't have ratings, so adding a dummy ratings value for analysis purpose.

add_star_if_needed <- function(x) {
  if (!grepl(" · ★", x)) {
    x <- sub(" · ", " · ★ 0 · ", x)
  }
  return(x)
}
```

```{r}
# Apply the function to the entire "name" column
airbnb_2$name <- lapply(airbnb_2$name, add_star_if_needed)
```

```{r}
airbnb_2 <- airbnb_2 |>
  separate(name, into = c("Type of House", "Ratings", "Bedroom", "Beds", "Baths"), sep = " · ")
airbnb_2
```

```{r}
airbnb_2 <- airbnb_2 |>
  separate(`Type of House`, into = c("Type of House", "City"), sep = " in ")
airbnb_2
```

```{r}
# Remove the star character from 'Ratings'
airbnb_2$Ratings <- gsub("★", "", airbnb_2$Ratings)
airbnb_2
```

```{r}
#0 meant no ratings to the house, for analysis purpose rather than keeping it 0 which can be possible rating, gave the value 10.

#Ratings are out of 5. So any but (0 to 5 ) value seems reasonable. 

# Replace " 0" with 10 in the Ratings column (no ratings)
airbnb_2$Ratings[airbnb_2$Ratings == " 0"] <- "10"


# Replace "New" with 11 in the Ratings column (new houses)
airbnb_2$Ratings[airbnb_2$Ratings == "New"] <- "11"


# Replace "studio" with 0 in airbnb_2$bedroom (no bedrooms/studio apts)
airbnb_2$Bedroom[airbnb_2$Bedroom == "Studio"] <- "0"

# Remove "bedroom" or "bedrooms" from the "Bedroom" column 
airbnb_2$Bedroom <- gsub(c("bedroom", "bedrooms"), "", airbnb_2$Bedroom)
airbnb_2$Bedroom <- gsub("s", "", airbnb_2$Bedroom)

# Remove "bed" and "beds" from the "Beds" column
airbnb_2$Beds <- gsub(c("bed"), "", airbnb_2$Beds)
airbnb_2$Beds <- gsub(c("s"), "", airbnb_2$Beds)

```

```{r}
sum(is.na(airbnb_2$Baths))
```

```{r}
# Identify columns with NA values
cols_with_na <- colnames(airbnb_2)[colSums(is.na(airbnb_2)) > 0]
cols_with_na
```

```{r}
# Total NA cases
sum(is.na(airbnb_2$Beds))
sum(is.na(airbnb_2$Baths))
sum(is.na(airbnb_2$reviews_per_month))
```

```{r}
# Drop rows with NA values in "beds" and "bathrooms" columns
airbnb_2 <- airbnb_2[complete.cases(airbnb_2[, c("Beds", "Baths")]), ]
```

```{r}
# Total NA cases
sum(is.na(airbnb_2$Beds))
sum(is.na(airbnb_2$Baths))
sum(is.na(airbnb_2$reviews_per_month))
```

```{r}
# Check min and max values
summary(airbnb_2$reviews_per_month)
str(airbnb_2$reviews_per_month)
```

```{r}
# Replace NA which means no reviews with 100
airbnb_2$reviews_per_month[is.na(airbnb_2$reviews_per_month)] <- 100

```

```{r}
str(airbnb_2)
```

```{r}
columns_to_convert_factor <- c("Type of House", "City", "Baths", "room_type")

columns_to_convert_num <- c("Ratings", "Bedroom", "Beds")

# Convert the specified columns to factors using lapply
airbnb_2[columns_to_convert_factor] <- lapply(airbnb_2[columns_to_convert_factor], factor)


# Convert the specified columns to num using lapply
airbnb_2[columns_to_convert_num] <- lapply(airbnb_2[columns_to_convert_num], as.numeric)

```

```{r}
levels(airbnb_2$City)
```

```{r}

airbnb_2$City <- as.character(airbnb_2$City)
airbnb_2$City <- gsub("austin", "Austin", airbnb_2$City)
airbnb_2$City <- gsub("Austin ", "Austin", airbnb_2$City)
airbnb_2$City <- gsub("Austin Texas", "Austin", airbnb_2$City)
airbnb_2$City <- gsub("East Austin", "Austin", airbnb_2$City)

```

```{r}
airbnb_2$City <- as.factor(airbnb_2$City)
levels(airbnb_2$City )
```

```{r}
airbnb_2$City <- as.character(airbnb_2$City)
airbnb_2$City <- gsub("AustinTexas", "Austin", airbnb_2$City)
airbnb_2$City <- gsub("West Lake Hills", "Westlake Hills", airbnb_2$City)
airbnb_2$City <- gsub("Westlake Hills, Austin", "Westlake Hills", airbnb_2$City)
airbnb_2$City <- gsub("Texas", "Westlake Hills", airbnb_2$City)

```

```{r}
airbnb_2$City <- as.factor(airbnb_2$City)
levels(airbnb_2$City )
```

```{r}
table(airbnb_2$City)

```

```{r}
table(airbnb_2$`Type of House`)
```

```{r}
table(airbnb_2$Baths)
```

```{r}
table(airbnb_2$room_type)
```

```{r}
# Convert the character dates to Date objects
airbnb_2$last_review <- as.Date(airbnb_2$last_review, format = "%Y-%m-%d")
```

```{r}
str(airbnb_2)
airbnb_2
```

-   There are 39 types of houses.
-   There are 21 Locations in Texas.
-   There are 36 types of Baths.
-   There are 4 room types.

### Exploratory data analysis, data visualization, and statistical analysis.

```{r}
#Univariate Analysis: Examining each variable individually.
# Checking for Outliers in price

#install.packages("gridExtra")
library(gridExtra)


  ggplot(data = airbnb_2, aes(x = price)) +
  geom_histogram(binwidth = 400, fill = "skyblue", color = "black") +
  labs(x = "Price", y = "Frequency") +
  theme_minimal()

```

-   This plot shows there are some extreme outliers present in price making it rightly skewed. This can be luxurious home, or price might be a human error. It makes it different from rest of the cluster. Therefore, not considering outliers for this analysis.

```{r}
# Removing the Outliers in the price
# Calculate the IQR for the 'price' variable

Q1 <- quantile(airbnb_2$price, 0.25)
Q3 <- quantile(airbnb_2$price, 0.75)
IQR <- Q3 - Q1

# Define the lower and upper bounds for identifying outliers
lower_bound <- Q1 - 1.5 * IQR
upper_bound <- Q3 + 1.5 * IQR

# Identify outliers
outliers <- airbnb_2$price < lower_bound | airbnb_2$price > upper_bound

# Create a new dataset without outliers
airbnb_no_outliers <- airbnb_2[!outliers, ]

# Check the dimensions of the new dataset
dim(airbnb_no_outliers)
```

```{r}
# Summary Statistics:
summary(airbnb_no_outliers)
```

```{r}
# Number of Airbnb properties under each host_id's.

filtered <- airbnb_no_outliers%>%
  group_by(`host_id`) %>%
  summarize(count = n()) %>%
  arrange(desc(count))
filtered
```

-   107434423 host_id has most of the houses listed on Airbnb.

```{r}
# Ratings given to Airbnb based on days of the week. 

# Using the mutate function to add the day of the week column to dataset
airbnb_no_outliers <- airbnb_no_outliers %>% 
  mutate(day_of_week = weekdays(airbnb_no_outliers$last_review))

# ratings count by day of the week
result <- airbnb_no_outliers %>%
  group_by(day_of_week) %>%
  summarise(count = n()) %>%
  arrange(desc(count))

result
```

-   Sunday is the day of week when people rate the Airbnb.

-   After that it gradually reduces during working days.

-   NA's represent no date present in the dataset. It can be new property, no ratings.

```{r}
# Top rated Hosts
ratings_value <- subset(airbnb_no_outliers,select= c(host_name, Ratings, City,`Type of House`,number_of_reviews_ltm ))

ratings_values_true <- ratings_value %>%
  filter(Ratings >= 0 & Ratings <= 5 & !(Ratings %in% c(10, 11))) %>%
  arrange(desc(Ratings))
ratings_values_true

# filtering out hosts with ratings 5/5
count_of_hosts_with_rating_5 <- ratings_values_true %>%
  filter(Ratings == 5) %>%
  summarise(count = n())

count_of_hosts_with_rating_5

```

-   Top 10 hosts are based in Austin area with ratings 5/5.

-   There are 2292 hosts with 5/5 ratings.

```{r}
# Most popular type of house
popular_types <- ratings_values_true %>%
                       group_by(`Type of House`) %>%
                             summarise(count = n()) %>%
                                  arrange(desc(count))
popular_types
```

-   " Home" is most popular AirBnB type of house in the Texas. Followed by Rental units and Condos.

```{r}
# Price based on Room Type vs Type of House

ggplot(airbnb_no_outliers, aes(x = room_type, y = `Type of House`, fill = price)) +
  geom_tile() +
  labs(title = "Room Type vs Type of House Based on Price") +
  xlab("Room Type") +
  ylab("Type of House") +
  scale_fill_gradientn(colors = c("yellow", "darkgreen", "black"), values = scales::rescale(c(0, 500, max(airbnb_no_outliers$price)))) +
  theme_minimal()

```

-   "Entire home/apt" listings tend to have higher prices, "Shared room" listings are generally the most budget-friendly option, and their prices are similar across different types of houses.

-   In Entire home/apt , "Place to stay" type of house is priced highest, followed by "Home", "Aparthotel" and "Bus", "Barn", "Casa Particular" have lowest prices.

-   In Hotel room, prices are moderate for "resort".

-   In Private room type, "bungalows" are priced highest, "home" and "guesthouse" are prices lower.

-   In shared room, only "Camper/RV" and "Townhouses" are costlier , rest are budget friendly.

```{r}
# Price based on Room Type vs Location

ggplot(airbnb_no_outliers, aes(x = room_type, y = City, fill = price)) +
  geom_tile() +
  labs(title = "Room Type vs location Based on Price") +
  xlab("Room Type") +
  ylab("Location") +
  scale_fill_gradientn(colors = c("yellow", "darkgreen", "black"), values = scales::rescale(c(0, 500, max(airbnb_no_outliers$price)))) +
  theme_minimal()

```

-   Room Type Impact: "Entire home/apartment" listings tend to be more expensive (darker tiles) than rest of the others listings in all locations."Shared room" tend to be budget friendly.

-   Location Impact: Bee Cave location is having higher prices, followed by Austin and WestLake Hills. However, if used shared room in austin that tends to be budget friendly, followed by Manchaca location.

```{r}
# Price based on Baths vs Bedrooms

ggplot(airbnb_no_outliers, aes(x = Baths, y = Bedroom, fill = price)) +
  geom_tile() +
  labs(title = "Baths vs Bedroom Based on Price") +
  xlab("Baths") +
  ylab("Bedroom") +
  scale_fill_gradientn(colors = c("yellow", "darkgreen", "black"), values = scales::rescale(c(0, 500, max(airbnb_no_outliers$price)))) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))
```

-   The price is higher for bathrooms in range of 3 to 5.5 and bedrooms in range 4 to 8.

-   Rest of types of bedrooms and baths have lesser price.

```{r}
# Price based on Ratings vs Number of Reviews

ggplot(airbnb_no_outliers, aes(x = Ratings, y = number_of_reviews, color = price)) +
  geom_point(size = 2) +
  labs(title = "Ratings vs Number of Reviews by Price") +
  xlab("Ratings") +
  ylab("Number of Reviews") +
  theme_minimal() +
  scale_color_gradient(low = "blue", high = "red") +
  xlim(0, 5) +
  ylim(0, 600)


```

-   The scatter plot appears to show a scattered distribution of data points, which suggests that there isn't a strong, linear correlation between ratings, number of reviews and price. Listings with different ratings and number of reviews can be found at various price points.

-   Outliers: There might be some outliers, which are listings with high prices but varying ratings, number of ratings. These outliers could indicate unique, premium listings.

-   Clusters: While there isn't a clear linear trend, we might notice some loose clusters or patterns. For example,many highly-rated listings are concentrated in a specific price range, or that low-rated listings are often more budget-friendly. However, these clusters are not as pronounced.

```{r}
# Price based on longitude vs latitude 

ggplot(airbnb_no_outliers, aes(x = longitude, y = latitude, color = price)) +
  geom_point() +
  labs(title = "Price Variations Across Latitude and Longitude Coordinates") +
  scale_color_gradient(low = "blue", high = "red") +
  theme_minimal()


```

-   From the above plot we can see that there is no strong correlation of latitude, Longitude with prices. Although we can see that at longitude greater than -97.8 there are some lesser price airbnbs along with some higher price ones.

```{r}
# Neighborhood vs Mean Price
airbnb_no_outliers_summary <- airbnb_no_outliers %>%
  group_by(neighbourhood) %>%
  summarize(mean_price = mean(price),number_of_listings = n())

ggplot(airbnb_no_outliers_summary, aes(x = neighbourhood, y = mean_price)) +
  geom_bar(stat = "identity", fill = "skyblue", color = "black") +
  labs(title = "Neighborhood vs Mean Price") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  theme_minimal()


# Sort the summary data by mean price in descending order
sorted_summary <- airbnb_no_outliers_summary[order(airbnb_no_outliers_summary$mean_price, decreasing = TRUE), ]

# Extract the neighborhood with the highest mean price
highest_mean_price_neighborhood <- sorted_summary$neighbourhood[1]
highest_mean_price <- sorted_summary$mean_price[1]

# Extract the neighborhood with the lowest mean price
lowest_mean_price_neighborhood <- sorted_summary$neighbourhood[nrow(sorted_summary)]
lowest_mean_price <- sorted_summary$mean_price[nrow(sorted_summary)]

cat("Neighborhood with the highest mean price:", highest_mean_price_neighborhood, "($", highest_mean_price, ")\n")
cat("Neighborhood with the lowest mean price:", lowest_mean_price_neighborhood, "($", lowest_mean_price, ")\n")

```

-   On x-axis, each bar corresponds to a different neighbourhood.

-   On y-axis, the mean_price indicates average price range for Airbnb listings that visitors can expect to encounter when booking accommodations in that neighbourhood.

-   By analyzing the number of listings and prices for each neighborhood, we can observe the price variation across different neighbourhoods.Some neighbourhoods have expensive listings while others are more affordable.

-   Neighborhood with the highest mean price: 78712 (\$ 500 )

-   Neighborhood with the lowest mean price: 78719 (\$ 113.8421 )