Visualizations that depict a consumer's utilization of domestic appliances aid them in effectively thinking about energy conservation. Energy disaggregation aims to break down the total power consumed into the amount consumed by each individual appliance through the use of a single smart meter. This estimation technique provides detailed information about energy consumption and is cheaper than techniques that directly measure each appliance. In order to improve the accuracy of disaggregation, we utilize the consumer feedback information without placing any special burden on the consumer. We propose a semi-supervised shift-invariant weighted non-negative matrix factorization method with auxiliary feedback that records the on or off status of each individual appliance. The experimental results obtained by applying our proposed method to household datasets show that our proposed method and the associated auxiliary information generated contribute to the improvement of the disaggregation accuracy.