A limitation of Chi Merge is that it cannot be used to discretize data for unsupervised learning (clustering) tasks [11]. Also, Chi Merge is only attempting to discover first order (single attribute) correlations, thus might not perform correctly when there is a second-order correlation without a corresponding first-order correlation, which might happen if an attribute only correlates in the presence of some other condition. Another shortcoming of Chi Merge is its lack of global evaluation. When deciding which intervals to merge, the algorithm only examines adjacent intervals, ignoring other surrounding intervals. Because of this restricted local analysis, it is possible that the formation of a large, relatively uniform interval could be prevented by an unlikely run of examples within it. One possible fix is to apply the x2 test to three or more intervals at a time. The x2 formula is easily extended, by adjusting the value of the parameter m in the x2 calculation.