Qualitative use of forecast variables in hybrid load forecasting techniques

Hybrid forecasting is a blend of various forecasting techniques. One such forecasting technique was presented by Rahman and Shrestha (1991) which utilized the attractive features of both the statistical and expert system based methods. The priority vector based load forecasting technique uses pairwise comparisons to extract relationships from pre-sorted historical hourly load and weather records for up to two years. The pre-sorting is done to identify seasonal boundaries and to categorize the day types (weekdays, weekends, holidays, etc.). The technique is adaptive in the sense that it internally generates the coefficients for relative influence of relevant variables (i.e., weather parameters) on the load. As these relationships change over time, such coefficients are automatically updated. This paper extends the technique by investigating qualitative use of continuous variable (e.g. temperature) as a means to overcome some of the difficulties confronted during the implementation when record high or record low values of these variables are encountered. This technique has been applied to forecast the hourly loads for a week, in summer when record high temperatures were observed, using 168-hour lead time. Results obtained by implementing the technique by using temperature as both (i) continuous variable, and (ii) qualitative variable, using the same set of historical data from a utility company are presented. Most, forecast errors are below 5% and many of the large errors are reduced by qualitative treatment of the variable temperature