Data Scientist | March 2020

Company: TravelNest

Position: Data Scientist

Location: NA

Level: nan

Outcome: Onsite Rejected

How was the interview process? What was it like?

There were two on-site interviews, the first behavioural and second technical. In the first interview, I found the smartphone app question quite difficult as I couldn't come up with something on the spot. The technical interview was straightforward as I was presenting my own work and my approach to the problem they asked me beforehand. Overall, I found that part okay, but at the end I was asked more specifically about how I would go about testing the pricing prediction product, and although I mentioned A/B testing I didn't know much about experimental design so wasn't really sure how I would go about it. 

I was particularly taken aback by the simple mean vs median question - although I would be able to explain it very easily without any pressure, in the interview I froze and couldn't offer a very simple explanation. I received feedback from the company saying that I had a lot of potential, but I needed to start from more basic descriptive statistics when going about new problems, keep to first principles. They also said I struggled to explain to a non-technical audience.

What technical questions were asked?

Business Case and Estimation

What was one of your solutions?

1. Discussed work I did on my thesis project analysing sensor data. Gave an explanation the results and limitations of the project.

2. I discussed the various types of data - customer, rental property, property owner, geographical, event (weather, festival, concert) data. I then went on to discuss how these various data inputs might impact the pricing of a rental property. I discussed how I would go about generating an occupancy rate prediction given property data and a certain price. From there I said we could use a grid optimization method to find the best price. I then went on to discuss how we could offer a price optimization tool to the customers (those renting out properties). 

3. I mentioned that the mean takes outliers into account much more, where the median is the middle of the data. I also mentioned that the median, when calculating e.g. number of rooms would be a whole number, whereas the mean would be fractional.

4. I said that I would improve my gmail app by integrating more things (I froze on this question and couldn't think of anything).