The Assignment

I was given the assignment to consider self-reported recommendation data for a product. The broad topic was to understand recommendation dynamics.

The data consisted of five variables:

• buyer_purchase_ts: Timestamp of the buyer's ("recommendee's") purchase
• buyer_id: Unique ID for each "buyer"
• recommender_id: Unique ID for each recommender
• buyer_trial_starts_ts: Timestamp for a buyers trial start
• recommender_trial_start_ts: Timestamp for a recommenders trial start, if available

I was also told that the standard pathway to a purchase is through a trial.

Questions

The assignment asked me to consider two questions:

1. Explore the data and present the three, in my opinion, most interesting findings
2. Present the three hypotheses I would like to explore next and explain why

I could present my solution in any way I deemed appropriate.

My Solution

My solution consisted of a presentation available here and a supplementary Jupyter notebook available here.

Remark

While the company was so generous to grant me permission to share this work, they asked that I remove any mentioning of the company's name. Thus, the name is either replaced with Company X or I speak of the product or the company.