Seattle Airbnb Data Analysis

Project Introduction

This project is a data analysis to Seattle Airbnb Open data, which describes the listing activity of homestays in Seattle.

Dataset

The dataset contains three files:

• listings.csv - including full descriptions, average review scores and price of 3818 homestays.
• reviews.csv - including 84849 reviewers' id and their detailed comments.
• calendar.csv - including price of 3723 homestays in a year from 2016/1/4 to 2017/1/4.

Target

When we travel to other cities, accomodation is the first thing we need to consider. According to the dataset and my life experience, I put forward following three questions, and the target of this project is to find answers of these questions.

1. When will it be cheaper to book a homestay in Seattle?
2. What are main factors that influence the price?
3. What factors people care about most when choosing an accomodation?

Method

To find the answer of each question, I used following methods to do data analysis.

1. Analyze calendar data, search the relationship between date and price.
2. Analyze listings data, search the relationship between price and some features we care about, and use LightGBM model to get importance of each feature.
3. Analyze reviews data, generate word cloud from reviewers' comments, to find the words which appear in comments frequently.

Result

In response to above questions , I analyzed the dataset and got the answer of each question as following.

Answer 1:

• It will be cheapest to book a homestay in January, while most expensive in July.

• Price always increase on Friday and Saturday, so try to avoid booking accomodataion on these two days in a week.

Answer 2:

• Location is an important factor that can influence the price. Average price in the region whose zipcode is 98134 is highest, while in the region whose zipcode is 98106 you can book cheap homestay. What's more, listings near Kaanapali are usually expensive while those near Holly park can be very cheap.

• The number of amenities listings have can also influence the price. In general, listings with more amenities usually are more expensive. But this phenomenon is not that unconditional, actually it also depends on what kinds of amenities listings have.

• For some kinds of amenities, listings with them or without them may have different prices. A homestay with TV will obviously more expensive than that without, But a homestay with cats or dogs usually cheaper than that without cats of dog. (In following picture, '1' means with that amenity, '0' means without that)

• Location, cleaning fee, available time, reviews(including number of reviews per month and review scores), number of amenities might be top 5 important factors for the price.

Answer 3:

• Location, property of accomodation, comfort, cleanliness might be factors that guests care about most.

Conclusion

Therefore, according to data analysis above, there are some tips for gusts to find cheap accomodations and some tips for new hosts to improve competitiveness and get more bookings.

Tips for gusts:

• At different time the price can be very different, try to avoid travel Seattle between June and August, and avoid booking homestays on Friday and Saturday. If you want to visit Seattle with lowest price, booking on weekday in January might be the best choice.
• Try to avoid choosing homestays near Kaanapali or in the region whose zipcode is 98134. You may find cheap homestays near Holly park.
• Don't choose homestays with too much amenities, those with necessary amenities can just be ok.

Tips for hosts:

• Homestays near some popular places or with convenient transportation are usually more preffered.
• If your location is not that good, don't worry. Try your best to make your house clean and comfortable.
• Make sure that your house has necessary amenities, like TV and air conditioning. It would be better if there is a kitchen in your house.

Libraries

All libraries I used in this project are as following:

• numpy
• pandas
• matplotlib
• seaborn
• sklearn
• lightgbm
• time
• warnings

Description of each file

• Seattle_Airbnb_Open_Data - Dataset, including three files.

• listings.csv
• calendar.csv
• reviews.csv

• listings_data_cleaning.ipynb - code used to clean listings data.
• listings_data_analysis.ipynb - code used to analyze relationship between price and other features and preprocess data using listings data.
• listings_data_predict.ipynb - code used to make price predict model.
• calendar_data_analysis.ipynb - code used to analyze calendar data.
• reviews_data_analysis.ipynb - code used to analyze reviews data.

Acknowledgement

Data Source - Seattle Airbnb Open Data https://www.kaggle.com/airbnb/seattle