Imputation Method for Multidimensional Data

Quality of data is the primary factor that affects the level of insights that can be fetched from data. One of the elements that significantly affect data quality is missing values. In some cases, missing values are unavoidable, and finding a technique to impute such values is difficult. Therefore, it is required that they are taken care of appropriately to get more accurate results. Although a few imputation strategies are available, sometimes those techniques carry out poor results as there is not a single definite framework that consistently provides the correct answer. This study proposes imputation method for multidimensional data. We have used a dataset containing records of various categories of crimes against women to demonstrate and evaluate the proposed data imputation method. The study also describes how correlated classes and the available data can be used to handle missing values using simple mathematical operations. The proposed approach provides guidelines to perform imputation for any multidimensional data.