2024-02-07
The Airbnb platform has revolutionized the hospitality industry by enabling individuals to rent out their properties to travelers, providing an alternative to traditional hotels.
As Airbnb continues to grow in popularity, understanding the characteristics and trends of Airbnb listings becomes increasingly important for hosts, guests, and stakeholders in the tourism industry.
Can we create a model to predict the success of an Airbnb listing based on its listing attributes?
In this project, we aim to analyze a dataset of Airbnb listings in a specific city to gain insights into the following aspects:
Data was collected from 6949 listings in Santa Clara region. The dataset was downloaded from Inside Airbnb.
Inside Airbnb collects and analyzes data from the Airbnb platform to provide insights into its impact on various cities around the world.
The website hosts datasets containing information about Airbnb listings, hosts, and reviews for numerous cities globally.
These datasets are typically obtained by scraping data from the Airbnb website and are made available for download in various formats, such as CSV files.
# Set the path to the Airbnb listings dataset file airbnb_path <- "/Users/jtruongjones/Downloads/Airbnb EDA/listings.csv" # Load the dataset into R airbnb_data <- read.csv(airbnb_path, stringsAsFactors = FALSE) # Check for missing values missing_values <- colSums(is.na(airbnb_data)) # Remove duplicate rows airbnb_data <- unique(airbnb_data)
# Extract number of beds from the "name" category using regular expressions
airbnb_data$beds <- as.numeric(gsub("\\D", "", gsub(".*\\b(\\d+\\s*bed).*", "\\1", airbnb_data$name)))
# Display the first few rows of the updated dataset
head(airbnb_data[, c("beds")])
## [1] 1 1 1 3 1 1
# Extract star ratings from the "name" column
airbnb_data$star_rating <- as.numeric(gsub("[^0-9.]", "", gsub(".*★(\\d+\\.\\d+).*", "\\1", airbnb_data$name)))
# Display the first few rows of the updated dataset
head(airbnb_data[, c("star_rating")])
## [1] 4.81 4.50 4.87 4.89 4.87 4.93
The two primary techniques used to preprocess included:
# Load necessary library for encoding
library(caret)
# Identify categorical variables
categorical_variables <- c("room_type", "host_name")
# One-hot encode room_type
encoded_room_type <- model.matrix(~ room_type - 1, data = airbnb_data)
# Label encode host_name
encoded_host_name <- as.integer(factor(airbnb_data$host_name))
# Replace original column with encoded one
airbnb_data$room_type <- NULL
airbnb_data$host_name <- encoded_host_name
# Concatenate encoded columns
encoded_data <- cbind(airbnb_data, encoded_room_type)
# Display the first few rows of the encoded dataset
head(encoded_data)
## id ## 1 4952 ## 2 11464 ## 3 21373 ## 4 62799 ## 5 75284 ## 6 106365 ## name ## 1 Place to stay in Palo Alto · ★4.81 · 1 bedroom · 1 bed · 2 shared baths ## 2 Rental unit in Santa Clara · ★4.50 · Studio · 1 bed · 1 bath ## 3 Place to stay in Palo Alto · ★4.87 · 1 bedroom · 1 bed · 2 shared baths ## 4 Place to stay in Palo Alto · ★4.89 · 1 bedroom · 3 beds · 2 shared baths ## 5 Place to stay in Palo Alto · ★4.87 · 1 bedroom · 1 bed · 2 shared baths ## 6 Guesthouse in Cupertino · ★4.93 · 1 bedroom · 1 bed · 1 bath ## host_id host_name neighbourhood_group neighbourhood latitude longitude price ## 1 7054 1127 NA Palo Alto 37.43932 -122.1574 65 ## 2 42458 493 NA Santa Clara 37.34415 -121.9870 94 ## 3 7054 1127 NA Palo Alto 37.43972 -122.1553 67 ## 4 7054 1127 NA Palo Alto 37.43934 -122.1572 83 ## 5 7054 1127 NA Palo Alto 37.43923 -122.1574 70 ## 6 551319 812 NA Cupertino 37.32194 -122.0051 289 ## minimum_nights number_of_reviews last_review reviews_per_month ## 1 7 84 2023-10-28 0.48 ## 2 3 20 2023-08-05 0.17 ## 3 7 266 2023-09-16 1.61 ## 4 7 157 2023-12-16 1.35 ## 5 7 214 2023-12-11 1.39 ## 6 30 187 2023-08-12 1.24 ## calculated_host_listings_count availability_365 number_of_reviews_ltm license ## 1 5 253 7 ## 2 13 152 4 ## 3 5 300 5 ## 4 5 309 6 ## 5 5 302 6 ## 6 1 365 2 Exempt ## beds star_rating room_typeEntire home/apt room_typePrivate room ## 1 1 4.81 0 1 ## 2 1 4.50 1 0 ## 3 1 4.87 0 1 ## 4 3 4.89 0 1 ## 5 1 4.87 0 1 ## 6 1 4.93 1 0 ## room_typeShared room ## 1 0 ## 2 0 ## 3 0 ## 4 0 ## 5 0 ## 6 0
library(scales)
# Identify numeric variables for scaling
numeric_variables <- c("price", "minimum_nights", "number_of_reviews",
"reviews_per_month",
"calculated_host_listings_count",
"availability_365", "number_of_reviews_ltm")
# Perform Min-Max scaling for each numeric variable
scaled_data <- airbnb_data
scaled_data[, numeric_variables] <- lapply(airbnb_data[, numeric_variables],
rescale)
# Display the first few rows of the scaled dataset
head(scaled_data)
## id ## 1 4952 ## 2 11464 ## 3 21373 ## 4 62799 ## 5 75284 ## 6 106365 ## name ## 1 Place to stay in Palo Alto · ★4.81 · 1 bedroom · 1 bed · 2 shared baths ## 2 Rental unit in Santa Clara · ★4.50 · Studio · 1 bed · 1 bath ## 3 Place to stay in Palo Alto · ★4.87 · 1 bedroom · 1 bed · 2 shared baths ## 4 Place to stay in Palo Alto · ★4.89 · 1 bedroom · 3 beds · 2 shared baths ## 5 Place to stay in Palo Alto · ★4.87 · 1 bedroom · 1 bed · 2 shared baths ## 6 Guesthouse in Cupertino · ★4.93 · 1 bedroom · 1 bed · 1 bath ## host_id host_name neighbourhood_group neighbourhood latitude longitude ## 1 7054 1127 NA Palo Alto 37.43932 -122.1574 ## 2 42458 493 NA Santa Clara 37.34415 -121.9870 ## 3 7054 1127 NA Palo Alto 37.43972 -122.1553 ## 4 7054 1127 NA Palo Alto 37.43934 -122.1572 ## 5 7054 1127 NA Palo Alto 37.43923 -122.1574 ## 6 551319 812 NA Cupertino 37.32194 -122.0051 ## price minimum_nights number_of_reviews last_review reviews_per_month ## 1 0.001375378 0.016483516 0.09847597 2023-10-28 0.03408267 ## 2 0.002100578 0.005494505 0.02344666 2023-08-05 0.01160261 ## 3 0.001425392 0.016483516 0.31184056 2023-09-16 0.11602611 ## 4 0.001825502 0.016483516 0.18405627 2023-12-16 0.09717186 ## 5 0.001500413 0.016483516 0.25087925 2023-12-11 0.10007252 ## 6 0.006976919 0.079670330 0.21922626 2023-08-12 0.08919507 ## calculated_host_listings_count availability_365 number_of_reviews_ltm license ## 1 0.008097166 0.6931507 0.04268293 ## 2 0.024291498 0.4164384 0.02439024 ## 3 0.008097166 0.8219178 0.03048780 ## 4 0.008097166 0.8465753 0.03658537 ## 5 0.008097166 0.8273973 0.03658537 ## 6 0.000000000 1.0000000 0.01219512 Exempt ## beds star_rating ## 1 1 4.81 ## 2 1 4.50 ## 3 1 4.87 ## 4 3 4.89 ## 5 1 4.87 ## 6 1 4.93
## price minimum_nights number_of_reviews reviews_per_month ## Min. : 10.0 Min. : 1.00 Min. : 0.0 Min. : 0.010 ## 1st Qu.: 72.0 1st Qu.: 1.00 1st Qu.: 1.0 1st Qu.: 0.260 ## Median : 125.0 Median : 2.00 Median : 7.0 Median : 0.710 ## Mean : 209.7 Mean : 11.07 Mean : 32.2 Mean : 1.195 ## 3rd Qu.: 199.0 3rd Qu.: 28.00 3rd Qu.: 33.0 3rd Qu.: 1.630 ## Max. :39999.0 Max. :365.00 Max. :853.0 Max. :13.800 ## NA's :607 NA's :1717 ## calculated_host_listings_count availability_365 number_of_reviews_ltm ## Min. : 1.0 Min. : 0.0 Min. : 0.000 ## 1st Qu.: 1.0 1st Qu.: 53.0 1st Qu.: 0.000 ## Median : 3.0 Median :179.0 Median : 2.000 ## Mean : 47.4 Mean :184.9 Mean : 7.363 ## 3rd Qu.: 14.0 3rd Qu.:330.0 3rd Qu.: 8.000 ## Max. :495.0 Max. :365.0 Max. :164.000 ##
Goal: Use linear regression and listing attributes
To predict Availability out of 365 year, and thus the success of an Airbnb Listing
## Root Mean Squared Error (RMSE): 130.7727
## ## Call: ## lm(formula = availability_365 ~ beds + price, data = train_data) ## ## Residuals: ## Min 1Q Median 3Q Max ## -223.894 -117.258 -0.065 137.448 166.861 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) 196.19689 3.17123 61.868 <2e-16 *** ## beds 1.59782 1.52304 1.049 0.294 ## price 0.01228 0.01660 0.740 0.459 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 130.5 on 4993 degrees of freedom ## Multiple R-squared: 0.0007628, Adjusted R-squared: 0.0003625 ## F-statistic: 1.906 on 2 and 4993 DF, p-value: 0.1488
Coefficients: If they are statistically significant (i.e., if their p-values are sufficiently low), it suggests that these variables have an impact on the availability.
R-squared: These values indicate the proportion of variance in the response variable (availability_365) explained by the predictor variables (beds and price). Higher values indicate a better fit of the model to the data.
Residuals: The differences between the observed and predicted values (residuals).
F-statistic: Assess the overall significance of the model using the F-statistic and its associated p-value. A low p-value suggests that the model as a whole is statistically significant.
Overall, based on our results, the model does not appear to be a good fit for predicting availability based on number of beds and price.
Ultimately, the dataset was missing some integral information for creating a prediction model. To improve the dataset, and create a better prediction model for investors it would be recommended to gather more data on