Challenges winner

The team led by Lautaro Borrovinsky, Data Scientist from AySA (Watershare Latin American Regional Hub) won Fiware4Water Challenge. he team defined a Machine Learning Model to take the Pearson’s correlation coefficient between lab data and sensor, along with the mean of the difference between lab data and sensor data and the standard deviation of the given property. It helps identify which sensors are working correctly. Afterwards the use of clustering techniques are used to differentiate between behaviours and not sliding windows due to data labs are not continuous and therefore data are quite irregular.

 

The integration with Fiware4Water was straight forward once the Machine Learning Model was created through the integration into the BentoML to provide an overview of the data validated subset to evaluate the result of the platform. The results of the challenge will help to make a simpler and more efficient detection of anomalies in the water supply which improve water quality without increasing costs. The assessment process involved jury members from the Milan challenge and independent entities with expertise in machine learning algorithms and standardization.