This paper presents the methodology followed by Broadcast Audience Research Council (BARC) India and Magic9 Media & Analytics Private Limited to develop a Viewer Attribution Model (VAM) to convert household level Return Path Data (RPD) into individual viewership data. The BARC currency panel data was used to attribute individual viewership to the RPD data. A conventional algorithmic model was developed using the principles of statistical matching and subsequently iteratively improved in order to produce better results. In parallel to the conventional algorithmic model, Machine Learning techniques were also explored. While the Machine Learning models did provide better accuracy for individual viewer classification than the algorithmic model, the algorithmic model produced better results for predicting the overall incidence of viewing. The algorithmic model was enhanced further by using two additional variables identified as important by the Machine Learning model. The final algorithmic model yielded a higher accuracy and performed better in predicting the incidence of viewing. This model also does reasonably well in the attribution of minutes for the top channels. A few areas for improvement have been identified and will be taken up in next phase of development in order to further enhance model accuracy.