MODE BEHAVIOUR IN RELATION TO BIN SIZE AND DATA DISTRIBUTION (Published)
The “mode” has been proposed as an appropriate statistic to improve estimate especially in situations when data distributions are skewed or contain outliers such as activity duration in project scheduling. Since the underlying distribution of activity duration may be unknown and different modes can be obtained using different bin sizes of the histogram method, this paper,investigates the effect of varying histogram bin width and data distribution on the behaviour of the mode. Random numbers were generated from five distributions commonly used to model project activity duration at five different levels and varying sample sizes. Each set of sample is then binned using varying histogram bin width, Sturges’rule and Scott’s rule. The grand mode for all levels per classification is recorded and analyzed. It was found that bin width does not significantly affect the behaviour of the mode but the value of the mode is significantly dependent on the data distribution and sample size.
Keywords: : Statistical mode, Bin size, Estimate, Statistical distribution, Uncertainty