Forecasting Software in Practice: Use, Satisfaction, and Performance
Abstract
Using survey data from 240 US corporations, we evaluated practitioners' use and satisfaction with forecasting software and its performance. Despite the many commercial forecasting soft-ware packages, only 10.8 percent of the respondents reported using them. Forty-eight percent reported using spreadsheets to make forecasts. Sixty percent reported being dissatisfied with forecasting software. However, we found that those who used commercial forecasting soft-ware packages had the best forecast performance, as measured by mean absolute percentage error (MAPE). Those using commercially available packages had errors 6.7 percent lower than those using spreadsheets and 17.2 percent lower than those who used no program. Also, they were more satisfied with their software than those using spreadsheets. In fact, users of forecasting software programs reported a 12.2 percent reduction in forecast error. We found that 61 percent of respondents routinely adjusted forecasts produced by software based on their judgment. Roughly 85 percent of respondents considered ease of use and easily understandable results the most important software features.