Modeling household oil consumption by using data mining

There are a lot of researchs made with HSB data set which contains characteristics of household and personal consumption of households. The aim of this research is to identify the factors on household characteristics which affect the consumption of olive oil, butter, margarine, other vegetable oils and consumable oils consumed by households and consumption situations of households with decision tree method by using HSB datas compiled by TÜİK. In the first part of the research, data mining concept and methodologies used in knowledge discovery are summarized. In the second part, informations about data mining modelling type; in the third part, generally the informations about fats and oils are given. And also, in this part, the factors affect the general household consumption, food consumption and the oil consumption expenditures are examined, and the studies in this field are scrutinized. In the execution part of the research, databinding and preprocessing phases of HBS datas compiled by TÜİK between the years 2009-2012 and net of income and expenditure datas of inflation were included first. With the data sets obtained, by using the clustering analysis and decision tree method which are data mining methods, the consumption profile in determined oil consumption cathegory of Turkey household is profiled. In consequence of the research, factors affect the consumption of four different kinds of oil consumed by the household are determined via decision-tree method. Despite the factors affect the consumption of fat kinds usually look alike, it shows differences on which fat kind is affected by which characteristic in which way. In the decision tree, those details are seen. While the most important characteristic on consumption of healthy, nutritious and expensive oils like olive oil and butter is the income, it's the size of the household on cheaper oils like margarines, other vegetable and consumable oils. When looking down the nodes and the details of a decision tree took place on each kinds of oil, it is seen that the factors which affect the consumption and the characteristics are changed, the details of these factors are examined in the research. Keywords: Household Budget Survey, Household Oil Consumption, Data Mining, Decision Trees