This project aims to predict missing sensor data as stated in the 2016 Harvard IACS Computational Challenge (http://computefest.seas.harvard.edu/student-challenge). Sensors are located as shown in the image below.

For our winning model, we used an inverse distance weighted averaging methode (see IDWmodel.py) which was then fed as input to a linear regression model(LR_model.py). Further improvement could be achieved using moving averages (see LR_MA_model.py / LR_MA_model_advanced.py) and exponential decaying (see LR_MA_model_advanced_weighting.py). Using optimized parameter settings, this lead us to victory!

(c) 2016 N.Drizard, L.Spiegelberg / Team Merkozy