<![CDATA[Abstract - In this paper, we propose a quadratic regression technique to predict types of products that consumer prefer to purchase based on his/her demographic background. The proposed technique is employed as a pre-processing for predicting the consumer spending behavior predictor system. The quadratic regression technique is developed based on the market survey data of the U.S. Bureau of Labour Statistics (BLS) data. The market survey data are sets of consumers that categorized in various demographics based on their spending on the various products. In the pre-processing unit, best customers (indexed spending) and biggest customers (market share) are used as the input and average spending as the output reference for every demographic information. In the experiment, the predicting result is compared between fuzzy logic technique and regression technique for pre-processing in fuzzy logic predictor. The experiment result shows that the prediction performance by using the quadratic regression technique is better against various techniques i.e. linear regression, purequadratic regression and fuzzy logic. Keywords: Quadratic Regression, Fuzzy Logic, Mean Squared Error (MSE), Ordinal Structure Fuzzy Logic (OSFL), U.S. Bureau of Labour Statistics (BLS), Consumer Behaviour Prediction]]>
Copyrights © 2015
