Real-time massive data analysis

Context summary

ThinkR was contacted in order to set up a tool to monitor the watertightness of a fleet of refrigeration installations. Refrigerant gases being powerful greenhouse gases, the challenge is to quickly detect leaks with the best possible sensitivity. The execution constraints were:

• real-time analysis
• a large volume of data
• the reliability of the model

Our intervention

• Data preparation: sampling, extraction
• Data processing: various approaches tested (Machine Learning and Modeling)

Result & added value

• Analysis report providing recommendations to better detect refrigerant loss