Profile Forecaster is a tool for forecasting time series. It can be used universally because it contains an open system of prediction methods. These are suitable for forecasting profiles of various origins. Thus we can forecast price development at the stock market, purchase or sale development, or any other time series stored in the database.
The module has integrated forecasting methods of ForecastX® and tailor-made methods developed by our company, namely
- Short-term electricity load forecast (for 1 – 7 days)
- Mid-term electricity load prediction (for 1 month to 1 year)
These special methods have been developed by MicroStep-HDO based on a requirement for forecasting the future development in electricity consumption.The forecast depends on the following regressors: date, temperature, exposure, season of year, holidays... All these regressors are considered in both the learning and the forecasted period.
Besides these periodically repeating regressors, our methods regard also the impact of extraordinary events that do not occur regularly. Such events are dealt with by so-called correction profiles which are used to adjust the input data at the beginning of the forecast and output data after the forecast of the regularized data has been made.