The project illustrates the cross-industry process for data mining ie. CRISP-DM using Seattle AirBNB dataset. The objective of the analysis and modeling of the dataset is to gain insights on the following -
- Listings growth over the years in various neighbourhoods of Seattle.
- Identify neighbourhoods : 2a. Favoured by the guests. 2b. Distinct neighbourhoods in terms of pricing and ratings.
- What times of the year,do we see availability at best pricings.
- Features that influence the property pricing.
The process structure for wrangling, modeling and deploying of the results is as below -
- Overview
- Business Understanding
- Data Understanding
- Data Preparation
- Data Modeling and Evaluation
- Deployment
This project requires Python 3.x and following Python libraries installed:
numPy
pandas
scikit-learn
matplotlib
xgboost
seaborn
statsmodels
calmap
Data files
-
calendar.csv 365 days availability schedule for all listings along with pricing.
-
listings.csv Main file which holds the listings data of the hosts.
Jupyter notebook code file for data analyis and modeling
Analysis of Seattle airbnb listings.ipynb
Intermittant files from analysis
abs_listings_final.csv
abs_listings_model.csv
Analysis Report
Analysis of Seattle airbnb listings.html
README.md
Results and insights have been collated as a blog. Objectives addressed from the analysis and modeling -
- Visualizing the linear relationship between Price vs review related columns and listings_count, we see that medium to low priced listings have garnered higher reviews with mostly high review ratings.
- Superhosts perform better in all parameters of review rating.
- Listings growth over the years
- Concentration of multiple listings in the neighbourhood groups.
- How are the neighbourhoods doing in terms of listing counts and review rating
- Price and review scores rating influence on neighbourhood groups
- Property available for rent for all 365 days /year across neighbourhoods in terms of listings count
- Availability based on review ratings across neighbourhoods
- Availability based on review ratings across neighbourhoods 10.Availabilty of property type and their pricing
- Availabilty and Price across the 4 quarters across neighbourhoods
https://www.sv-europe.com/crisp-dm-methodology/#evaluation
https://xgboost.readthedocs.io/en/latest/python/python_api.html
https://datascience.stackexchange.com/questions/34209/xgboost-quantifying-feature-importances
http://www.grroups.com/blog/an-information-gain-based-feature-ranking-function-for-xgboost
https://www.visitseattle.org/neighborhoods