A Comparative Study Of Two Knowledge Discovery Tools: Barchart Versus Scatterplot
This paper reports the results of an empirical examination of the effectiveness of two knowledge discovery tools (barchart and scatterplot) in the context of a sales forecasting task. The main results of the study indicate that both tools were reasonably suitable for well conveying associations among task variables and offering improvements in prediction accuracy when compared to a naive predictor. There is however, still much ground for improvement towards a theoretical optimal case. In addition, findings show that a scatterplot was significantly more beneficial than a barchart in enhancing forecasters’ knowledge and performance of the task. This superiority can be potentially explained in terms of the favorable aspects of Cartesian graphs and the greater concentration required for using the tool.