Data Conversion Process Framework to Generate Individual-Level Nutrition Data from Household-Level Grocery Data

This paper presents a data conversion process involving household grocery data. The household grocery data were gathered from the primary source which is directly from 50 selected household in Shah Alam for 5 consecutive months. The data transformation was done to convert the grocery data into the nutrition data. The converted nutrition data will be tested using data mining classification algorithms, and the patterns generated from it will be explored for obesity prediction purposes. In the data transformation process, the raw grocery data has undergone several data pre-processing and conversion methods. These processes have been done by the nutritionists as the knowledge on nutrition field are needed in performing this task. The processes involved are calorie conversion, macronutrient grouping, food pyramid grouping, and food categorization. There were five methods have been conducted to perform the conversion task which are food composition database, offline and online market survey, food pyramid and knowledge theory on nutrition. The conversion process has been gathered to form Data Conversion Process Framework. This paper also introduced the use of estimation formula using BMI weightage as a method to generate the individual-level nutrition data. The nutrition data generated from the grocery data processing and the conversion process using the BMI weightage highlight the significance of the study. The output from this study (nutrition data) will be used in the later stage of the study as the input data in the development of obesity prediction modelling.