In addition to being a provider of weather data and weather based solutions, Weather Source also operates a Weather Data Science Lab (“WDSL”) to conduct research on how weather impacts demand for various products and/or impacts businesses. In the WDSL we process demand data against many 1000’s of weather impact tests to identify and quantify the impact of weather on the demand in question. The result is a ranked set of weather impacts models that can predict demand based on weather inputs.
The standard approach in processing data through the WDSL includes:
- Acquire demand data (process needed legal/security requirements and acquire data);
- Demand data prep (successfully load into database and verify results);
- Demand data exploration (quantitatively understand data, look for questions, get answers);
- Configure tests (set date ranges, locations, weather parameters, etc.);
- Execute test suite;
- Evaluate results (steps 3, 4, 5 & 6 may iterate a few times);
- Produce final report for client.
The resulting final report provides a description of the top weather impacts along with statistical summaries of weather impact model performance (e.g. RMSE, MBE, etc).