What can be done to improve AirBnB?

I want to start this journey off with a few questions? How can I provide quality of life and knowledge changes to all parties involved with AirBnB? What further analysis can be done to potentially improve the user experience? What information could be provided to users to help them make better decisions when choosing a place to stay at? What potential improvements can be made to make analysis easier on the backend for those doing their own analysis or even for AirBnB analysts?

These were the questions on my mind when I was presented and started to dig through the AirBnB dataset for Seattle. I got straight to work by doing what every analyst does best, cleaning.

After cleaning and getting the dataset ready for analysis, I started to think of approaches to my questions.

My first approach was to create a price prediction tool. This tool can help both those that are putting listings on AirBnB and those looking for a place to stay. My tool will take in the information a listing provides and places a price suggestion. This tools has an accuracy of approximately 75%, which is actually quite high for this type of prediction tool.

This tool can be super useful to those making listings because they can get an easy estimate of what they should price their listing at to be competitive with other similar listings on the market.

For those that are trying to find a place, they can use this tool to give them more information about the current listing they are interested in. It will immediately let them know if they are over-paying or under-paying for that current listing and by how much. This will allow the user to make a more educated decision before renting a place.

The next thing that interested me was, looking at how rating affected the price of a listing. I was curious to see if as the ratings got better and better, how would the price be affected?

This plot above reveals the relationship between the rating a listing has and it’s price.

For starters, you can see the slight uptrend between rating and price. This means that the more a listing costs the better experience you will have at that place.

Again, this is true to an extent. We can see that people should avoid the low rating and cheaper listings. It tells us that people did not enjoy their time.

Based on this chat, we can see that if someone is willing to pay $400 and up they are almost guaranteed to enjoy their vacation while those that spend less can have much varied opinions on their vacation.

My last idea was to help those who want to analyze datasets provided by AirBnB or those that work at AirBnB itself. I created a tool that correctly predicts the type of room approximately 91% of the time.

The reason this can be very beneficial is because it can broaden the amount of data that can be analyzed. This tool can be used to fill in the gaps that can be missing.

Another potential use for this tool could be a sort of autofill when creating a listing. When a user starts to fill out the information regarding the listing they want to post, it could potentially fill in the type of room automatically.