The training data is pulled from raw logs. It has three columns that record user activity over time: user_id: A hash that uniquely identifies the user. activity_date: The date of the activity activity_type: The type of activity 6sense would like to predict which users will purchase given the user's activities. Using the training data, build a model that predicts which user_id's will make a purchase in the future. Use the model to score the test data from most likely to least likely to purchase. Please submit the following: (1) Descibe which activity types you believe are most useful in predicting which user will purchase in the future. (2) Provide a file containing the 1000 user_id's most likely to convert. (3) [optional] Any work used to generate the list and/or notes you'd like to include.