data.csv represents a single simulated dataset (out of the 500 generated for Scenario 1 in sec 5.1 of the main paper). Each row of the data represents a single ad or conversion event for a single user, so that the complete path data for a single user spans multiple rows. There are 3 columns:

- user_id: an integer identifier for the simulated user. Values go from 0 to 999,999. - event_type: A string taking one of three values: “conversion” indicating that the row is a conversion event, “ad_type_1” or “ad_type_2”, indicating that the row is an ad event. The two different types of ad events are as described in the paper. ad_type_1 has short, medium, and long term effects of 2, 1.5, and 1.2, respectively, while ad_type_2 has short, medium, and long term effects of 1.5, 1.2, and 1.0, respectively.
- time_in_days (double): The ad and conversion times are simulated as being in [0, 30], measured in days. This indicates when the simulated event occurred.
