EY-NEXT-WAVE-CHALLENGE-2019

• The objective was to gain insight on citizens move along with the smart city project. With data on trips recorded by the mean of devices, we were challenged to predict if a t rip ends into the city center given data on time, the entry and exit locations 2D coordinates, and speed measurements.

• Overall we were ranked 23/373 and 4/30 in France where we made it all the way to national finals.

• To understand our approach

  • you can refer to the pdf file Next_Wave_Presentation.pdf
  • All the helper functions can be found in the file helper. makedata.py and getfeatures.py implement both data preprocessing and features engineering. The models are coded in the file predictors.py.
  • To reproduce the result: (1) split first the data into train, eval and test with split_data.py, and (2) run the python script run_classification.py. Yet if you are interested in our regression model (i) just estimate the modes of the arrival coordinates with likelihood_modes.py and (ii) run the regression with run_regression.py.