Categorical Data Transformation Methods for Neural Networks

摘要

Data mining is the process of analyzing and exploring large dataset from different perspectives in order to extract hidden predictive and useful information -information that can be used to increase revenue, cut costs, or both. There is a need to pre-process the data to make it easier to mine for knowledge. Data preprocessing includes data cleaning, data transformation and data reduction. This study addresses data transformation, which transformed categorical data to numerical data. In this paper, we proposed a data transformation method, information probability, and used neural network to predict motor vehicle injury accident with information probability in traffic safety domain. Experimental results show the significant improvement achieved by the proposed method. Accurate results of such data analysis can be useful for traffic safety engineer or policy maker to set up preventive countermeasure.