nyc_airbnb_capstone

Overview

Airbnb is a marketing company that acts as a broker to provide short-term homestay experiences that offer several advantages like exposure to everyday life in another location, and the opportunity to experience local culture and traditions. Multiple reports forecast an increase in future Airbnb market capital and revenue with a growing economy.

Goals

Our client is a successful investment and venture capital firm based in New York, owns several successful businesses and now wants to invest in new york based Airbnbs but wants to understand in-depth knowledge and information about the market before strategic decision-making.

Data Specifications

This project will use open-source data data of the year 2019 from AirBnb.

The dataset of 48,896 rows contains information about location, room type, prices, reviews, availability, and host information.

Project Structure

nyc_airbnb_price_prediction_ml_project
1. Data_Wrangling New.ipynb
├─ 2_Exploratory_data_analysis_EDA.ipynb
├─ 3. Pre_processing_&_Training_Data_Development.ipynb
├─ 4. Data_Modeling.ipynb
├─ Data
AB_NYC_2019.csv
│  ├─ nyc_data_after_EDA.csv
│  └─ nyc_data_cleaned.csv
├─ LICENSE
├─ New_York_City_Map.png
├─ Plots
Continous Variable Relationship with Price Plot.png
│  ├─ Correlation Matrix.jpg
│  ├─ Mean Absolute Error_MSE.jpg
│  ├─ Mean Square Error_MSE.jpg
│  ├─ Neighbourhood Plot.png
│  ├─ Percentage Room Distribution1.png
│  ├─ Percentage Room Distribution2.png
│  ├─ Percentage Room Distribution3.png
│  ├─ Percentage Room Distribution4.png
│  ├─ Percentage Room Distribution5.png
│  ├─ R2_Score.jpg
│  ├─ Word Cloud.jpg
│  ├─ tree-out.png
│  └─ tree1.png
Presentation
│  └─ Capstone_Presentation_AirBnb_NewYork.pdf
├─ README.md
├─ Report
│  └─ Capstone_1_Report_NewYork_AirBnb.pdf
└─ testing.ipynb

New York Map

Neighbourhood

Continous Variable Relationship with Price

Correlation Matrix

Wordcloud

Metrics

Model Comparison

Model	Mean Squared Error	R2 Score	Mean Absolute Error
Linear Regression	0.22	40.58	0.17
Decision Tree	0.21	48.2	0.15
Bayesian Regression	0.22	40.58	0.17
Ridge Regression	0.22	40.58	0.17
Lasso Regression	0.22	40.58	0.17
Gradient Boosting Regression	0.19	57.31	0.14

We can see that the Gradient Boosting Regression model has the highest R2 score value and the least error. Hence it is the best model for our Airbnb project.