Utility provider runs 7M forecasts natively in the warehouse
- 7M Time series
- 2M Customers
- 10 Products
Forecasting at full scale, inside the platform
A leading multi-utility provider in Latin America needed to forecast energy, gas, and water consumption for two million customers across 10 distinct products. The algorithms existed. The expertise was there. What was missing: infrastructure capable of running the problem at its true size—without moving data outside the warehouse. By building the entire forecasting system natively within Teradata, the organization went from zero forecasting capability at scale to seven million production-grade time series running end to end, with no external processing required.
The models were ready. The infrastructure wasn’t.
Forecasting consumption across two million customers and 10 products generated approximately seven million time series—far too large to extract, transform, and process externally. Moving that volume out of the warehouse wasn't viable: the data was too large, refresh cycles too frequent, and extraction costs too high. Without a way to run the full problem natively, the organization had no production forecasting capability at this scale. It wasn't because the science was lacking, but because nothing the company had could handle it.
7 million time series, in database and in production
Teradata built the entire forecasting pipeline natively inside its platform—end to end, with no data movement and no external processing. Each time series is automatically classified by forecast policy. Model selection adapts dynamically to the maturity of each series, from shorter histories to longer, more complex patterns. Every forecast includes confidence intervals and a standard accuracy metric set, fully comparable across the entire portfolio. The six-month projections feed directly into a segmentation engine that scores the full customer base by current and potential value, separately for business to business (B2B) and business to consumer (B2C).
From no capability to full-scale intelligence
Where no production forecasting capability existed before, the organization now runs seven million time series in-database, continuously, at scale. Every product. Every customer. Measured consistently. The forecasts don't just predict demand—they feed a segmentation engine that identifies the highest-value B2B and B2C opportunities across the full customer base. This is what in-database analytics is for: not proof-of-concepts or downsampled experiments, but full-scale, production-grade intelligence running natively, governed, and ready to grow.
Learn more
接続しましょう
Teradata が、ビジネスの成果を加速し、必要なビジネスの俊敏性を提供するのにどのように役立つかをご覧ください。
当社の営業担当者がお手伝いします。