Median polish techniques for analysing paired data

Median polish is used as a data analysis technique for examining the significance of various factors in single or multi-way models. The main goal of this research is to analyze paired data using median polish technique in order to get information from the data such as difference between the paired data through column and row effects. Median polish algorithm is useful in removing any noise in the data by computing medians for various coordinates on the data set. In this research, we will focus on paired data. Some effects such as overall, rows and column were determined using median polish algorithm. In this research, one-way and two-way median polish has been expanded for pairwise scenario. The pairwise median polish criterion addresses the fairness of declaring a difference between a paired data. Paired data involves collection of data prior to the treatment and compare it with after the treatment. We extend the analysis of paired data for the case of missing values. Later, exercising of comparison values in a two-way median polish for paired data was implemented to verifying the association between rows and columns effects. In addition, to determine whether there is need for transformation of data or not. Pairwise median polish model is successfully employed in the analysing the comparison and verification of the difference between paired data of grain yields in classification of contingency table. For the two-way median polish for paired data, comparison values calculated shows there is no association between rows and columns effects and transformation of data is not required in this study. The median polish provides simple estimation of main effects for paired data as well as various factor effects. The findings also have shown that there is a difference in grand effects of both after treatment data without missing values and imputed values using paired median polish procedure.

Preview Text FS 2017 3 IR.pdf Download (1MB) | Preview

Actions (login required)

View Item