We study the information contained in the dataset K =[X|Z] using the statistical method called Correspondence Analysis (CA) as a multivariate graphical technique designed to explore relationships among categorical variables. A descriptive approach is applied on the whole dataset before we partition the dataset into two parts: the first part is data subset X (quantitative variables - namely, the budget of household expenditure) for which we focus on the explanation and prediction; the second part is denoted as data subset Z which consists of additional qualitative variables (such as gender, age, household income, level of education). Finally, we propose a method to analyze a budget table with external information in order to predict the expenses of goods and services knowing the global amount and the profile of new subject.