cometyang / axa Goto Github PK

##AXA - Driver Telematics Analysis

Train N random forest model for each driver in the data set.

• Rather than consider each trip as a single instance, divide the trip in equals part (default=4). Then sum up the probabilities at the very end for each trip.
• Create features with different window
• The model is train on all the driver trips plus five times more trips randomly picked from all other drivers.

Just run "python model.py" and you're good to go! You'll need first to unpack the driver.zip, create a /data and /submission folder in order not to fail the script.

• n_jobs: allow multiprocessing spawn N jobs
• use_cache: use previously preprocessed data file
• n_drivers: drivers to process and save to file
• windows: windows for calculate features (1,15,30,60 seconds)
• part: number of part to split a trip into
• n_quantiles: number of features to create
• size: if not None, split trip in equals size parts rather than equals parts