Platinum is a system that automatically improves the quality of Kpler’s total load and generation per fuel type datasets by filling gaps and correcting outlier values.
What it does
Platinum correction detects issues in raw data reported by any source, such as missing or abnormally high/low values. These are then corrected using in-sample predictions from our proprietary models.
If the original provider later updates or backfills correct data, Kpler replaces its corrections with the new source values. Corrections are applied historically across the full time series, up to and including D-1.
How it’s applied in the API
Platinum correction is currently applied to:
When querying these endpoints with provider=kpler, Kpler automatically returns the cleaned data.
To help track corrections, each response includes a boolean field: is_fixed_data: indicates whether each hourly value was corrected by Kpler. Here is a response example :
Should the provider backfill the correct data, our correction will be overwritten and is_fixed_data: will change to false.
Platinum corrections are applied wherever Kpler operates a forecast model.
Please refer to our forecast coverage here: