The Sensitivity of Machine Learning Techniques to Variations in Sample Size: A Comparative Analysis

A comparative analysis of the performance of some well-known classificationtechniques (Discriminant Analysis, Quinlan’s See5, and Neural Networks) and certain machinelearning systems of recent development (ARNI, FAN and SVM) is conducted. The chosenclassification task is the forecasting of the level of efficiency of Spanish commercial andindustrial companies. Assignment of the firms is made upon the basis of a set of financialratios, which make a high dimension feature space with low separability degree. In the presentresearch the effects on the accuracy of variations of each technique in the estimation samplesize are measured. The main results suggest that ARNI and See5 yield the best results, evenwith small sample sizes.