This is the source code of Team Data Preparator for the 1st LEC Data Challenge.
🥈 This solution reached the 2nd place in the competition.
You may want to check out the following:
🔍 🔍 🔍
The task was to detect faults in sensor measurements.
Our solution forecasts the sensor measurements and flags regions if the real values significantly deviate from the forecast. We train an ensemble of ARMAX models and combine their predictions with a weigthed median, where the weights are based on the Spearman's rank correlation coefficient of the sensors, along with a bit of post-processing filter magic.
Package requirements are specified in the requirements.txt file.
pip3 install -r requirements.txt