Outlier Detection on Mixed Type Data by Using AVF and Z-Score Algorithm

Outliers are objects that deviate from other objects in general. It needs special attention because it can interfere to the data distribution and make the discovery of important information from the data set become more difficult. Many methods have been developed to detect the outliers. However, these methods focus only on the categorical data or numerical data separately. In real life, facts are described by using mixed type data. In this paper, the outlier detection was carried out on mixed type data by using AVF algorithm and Z-Score algorithm. The Z-Score algorithm was used to detect outlier in the numerical data. The AVF algorithm was used to detect outlier in the categorical data. By combining the outlier rankings from  the AVF algorithm and the Z-Score algorithm, some outliers in the sales transaction data in UGM Pharmacy could be found. The detected outliers were peculiarities in the pattern of the sale transactions in UGM Pharmacy from the 1 st to 16 th of April 2016.