The ABC-XYZ analysis modified for data with outliers

摘要

The ABC-XYZ analysis is widespread in modern business, especially in the inventory control, marketing and logistics. It is used for classification of the assortment depending on the revenue or profit of sales of the enterprise according to the ABC-part of the method and the amount of sales for the XYZ-part. The paper proposes a modification of the ABC-XYZ analytical technique for the case of the presence outliers in the data. The modification of the method allows identification of the outlier by means of the Grubbs' test. After excluding the outliers, the mean and the standard deviation of the truncated sample are used to re-estimate average sales, recalculate revenue and coefficient of variation, regroup and finally help decrease costs by means of better forecasting and changing strategies of inventory management, e.g. choosing more appropriate replenishment systems for products and avoiding bullwhip effect in a supply chain. The new technique was applied to a simulating data set and demonstrated the results of classification that differ from the classical one. The suggested method must be considered as a preferable technique for classification as it is based on more reliable and accurate data and robust estimation of mean value and standard deviation.