Receiptlog applied to forecast of personal consumption

Abstract

A purpose of this research is to improve a user's balance of income and expenditure by forecasting and presenting future consumption from past purchase information. First, we recorded various purchase information by scanning a large number of receipts and applying them to OCR (Optical Character Recognition). We named this system “Receiptlog”. From data of Receiptlog, we found out cycles of time and dependence of place in users' consumption behavior. Using these features, we forecasted amount and possibility of future consumption. Next, we developed an application for iPhone, “Expenditure Forecast”, that presented future information to users like a weather forecast. We evaluated this system and found out that presentation of forecast had effects on users' consumption behavior.