JetFuel Takehome - for ML/AI Optimization candidates!
This is a optimization exercise that will give you the chance to see the types of problems we're working on here at JetFuel!
In particular, we're going to be predicting the click-through rates on ads. We've collected a training set of 1M rows of existing data, and a test set of 100k rows. You can access the data by unziping the archive 'data.zip'
Your task is to train an optimization model the predict the whether or not an ad will be clicked on or not.
In both files, the data is in the following format:
id -- ad identifier
click -- 0/1 for non-click/click
hour -- format is YYMMDDHH, so 14091123 means 23:00 on Sept. 11, 2014 UTC.
C1 -- anonymized categorical variable
banner_pos
site_id
site_domain
site_category
app_id
app_domain
app_category
device_id
device_ip
device_model
device_type
device_conn_type
C14-C21 -- anonymized categorical variables\
Feel free to include multiple models in your submission. We'd like to see your full process, so please include models you trained that didn't work out!
Download the
cd optimization-takehome/
unzip data.zip
open predict.py