Uber Performance Analysis

Project Details

Using the dataset provided from StrataScratch, we want to understand what factors best predict whether a Uber signup will start to drive. We did data cleansing and analysis to determine and finalize which factors were important, what recruitment method has the most people completing a drive, the percentage of drivers completing the first trip, total signups by signup channel and city, and more.

Accomplishments that we're proud of

Found three main cities, each separated to look into the different signups which are relevant to whether drivers completed their first ride. This is achieved by using a correlation heat matrix to see what variables impacted others and looked into training and testing the significance of signups and cities toward completing first drive.

We were also able to conclude that almost 76% of signups did not complete and we wanted to know what could be the cause and how can this improve. Most people in all three cities signed up because they were paid and for those who have completed their first drive, most signed up because of referrals.

Challenges we ran into

Looking through the dataset, we found that there was a significant amount of missing values which decreased the dataset. However, we used these missing values to our advantage by changing the values to 0. We then focused more on the signup channel and whether drivers completed their first trip. This is attained through general trained signups and comparing whether the drivers completed their first trip. To further improve on this, we did a more specific model, which is Decision Tree, that looks into the specific cities and referral rates to test whether the drivers have completed their first trip.

What's Next?

We have several ideas on how Uber can implement next steps to improve getting more first ride trips; offering time-sensitive discounts to engage more trip orders, promote more Uber marketings and ask for engagement on areas with low first-trip rates, drivers with no first trip should be considered first priority to be offered when new rides are shown up, and Uber giving out first trip discount to all users, to build trust and connection, in order for customers to order even more rides in the future. We believe that with these improvements people would start doing first trips in the near future.