List the top three most important variables Uber could use to estimate the ETA for passenger pickup, beyond the ETA on Google Maps.

Approach

For any question that asks to list the "top X" answers, it's generally worthwhile to go through the "broad, then deep" strategy. Essentially, you'll want to enumerate as many possible valid answers as you can think of, and then stack-rank them according to how they align with the goals of the question. In this case, we'll want to think of a variety of possible variables that might affect ETA outside of the Google Maps ETA. A good approach to devise possibilities is to think of what causes you to be later or earlier than your navigation's ETA.

Variables to consider

• Driver speed rate. Perhaps certain drivers drive faster than others.
• Current driver speed. As the driver accepts a new passenger request, the driver may be going 65mph on a highway, which could prevent the driver from easily switching lanes and moving on course to pick up the passenger.
• Finding the passenger. It's generally difficult for Uber drivers and passengers to connect for pickup time. There are potential greater difficulties picking up a passenger at busy locations like a concert venue.
• Route wrong turn rate. Given a route, how likely is it that drivers have issues with correctly following the route?
• Weather condition. Poor weather conditions can cause delays in arrival time.

Analyzing the variables

After coming up with a reasonable list of potential variables, let's now analyze each one and consider its importance in calculating the ETA.

• Average driver speed. The variance here is likely to be fairly low - most drivers will probably be driving close to the speed limit. Even if some drivers reliably exceed the speed limit and beat the ETA, the illegal behavior is not something Uber would want to encourage in its ETA calculation.
• Current driver speed. Again, this is an interesting observation, but isn't significant. The probability of a driver on the highway receiving a passenger request is low. It's more likely that the driver will have just dropped off a passenger.
• Finding the passenger. This is a great variable to consider - it's actually fairly likely that an Uber driver is picking up or dropping off a passenger in a very populated concert venue or festival. It may be particularly challenging to connect with the driver in this crowded setting. Let's keep this one as an option.
• Wrong turn rate. This one seems like a great choice. There are a combination of variables that can estimate the likelihood that a driver would make wrong turns on the route, including the driver's wrong turn history and the route's wrong turn history. Let's keep this one as an option.
• Weather condition. This one is another great choice. It is substantially more challenging to navigate a route in poor conditions. Let's keep this one as an option.

Ranking the variables

We now have three top variables from our list that we think are relevant for Uber's ETA calculation. Now, let's try to be a bit more precise and recapitulate our points.

1. Uber should add extra time related to the crowdedness of a location, especially at concert and festival venues.

2. Uber should add extra time related to the likelihood of wrong turns, and the costliness of these wrong turns, based on historical data.

3. Uber should add extra time given poor weather conditions.