Effects of Heavy Tailed Distribution on Statistical and Neural Network Based Control Charts

Abstract

Artificial Neural Networks (ANN) had been used for the detection and classification of patterns in control charts. It has been shown that neural network can detect smaller shifts better than statistical control charts. However, nearly all studies in this area assume that the in-control process data in the control charts follow a normal distribution. In our study, we focus on the effects of heavy tailed distributions on the performance of neural network based control chart and statistical control charts. Statistical control charts like Shewhart X control chart, Exponentially Weighted Moving Average (EWMA) control chart and Cumulative Sum (CUSUM) control chart are presented to make the comparison of the effects of heavy tailed distribution with Neural network based control chart. The criterion to compare the performance of both types of control charts is the average run length (ARL). From the results, the neural network is less robust than the statistical based control charts in the presence of heavy tailed data.