As the weather has got much worse, I haven't been to be riding my motorcycle as much, and we discussed buying a second car, so we don't have to deal with all the hassle of sharing a single vehicle.

I came across a site called; it's essentially Airbnb for cars. Helping to offset some of the cost of car ownership, one can list their car for others to rent out.

I liked the idea of this, and decided to do some research into what sort of car I should buy, I reached out to the company to try and get access to some of the data, but without much luck. Instead, I worked out that all their API calls are public, and decided to get to work.

I did all this in a Jupyter notebook on Google Colab:

Using my local postcodes to my areas, I was able to replicate a search on the site, and pull down all the listings with the following information on each of the cars:

  • availability_text
  • available
  • body_type
  • costs
  • distance
  • img
  • key_handover
  • lat
  • lng
  • make
  • model
  • negative_review_count
  • number_of_seats
  • parking_type
  • positive_review_count
  • street
  • suburb
  • total_trips
  • transmission
  • url
  • vehicle_name
  • year

I pushed all this data into a pandas dataframe, and then created a new column estimating the total income based on a number of trips, and vehicle price (per hour and per km). The income number is likely very wrong but is relative, so still very useful.

After looking at just my local postcodes, I realised that my sample size was much too small, so I decided to get all data in Melbourne and Sydney postcodes.

Here is what it looks like grouping by car type, with total income from trips (approximate) on the horizontal axis, and dot colours indicating make:

What size car makes the most money?

Car by body type vs income (color coded by make)

This looked good, and from here I could see that the best option is a small car or a ute (as I was not interested in a van, they are pretty expensive).

I like the idea of a small car more, seems like a lesser risk, and easier to park, run, etc. I am open to a manual, but wanted to know of the transmission would impact income:

What make of small car and what transmission?

Small cars by make vs income (color coded by transmission)

Looking at this graph it seemed pretty clear that a Volkswagen, Mazda, Toyota, Kia or Hyundai with an Automatic transmission was the way to go.

Next, we wanted to know, just out of curiosity if we should put the car in my name, or if it's better to be in my girlfriend's name. So we decided to look into the gender of the owners of the car. The package I used was called gender. Lots of assumptions here, firstly binary gender only, and names like Alex and other fairly genderless names it didn't do well, but for the sake of generalisation it does ok:

Car listings by gender vs income (Color coded by make)

Looking at this, it was interesting to see that a number of the most successful listing happened to be what was considered girls names.

All in all, this was a pretty fun exploration and was insightful for when/if we decide to purchase a second car.