In this work, we develop prediction models using regression, artificial neural network (ANN), and linear programming (LP) for forecasting voter turnout in elections. This study attempts to propose a novel method of using linear program for prediction modeling. We consider the case of Tamil Nadu Assembly Elections and use the various socio-economic features and demographics as input to build our model. The metrics used for evaluating the performance of forecast are Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). We also report the Maximum Absolute Error (Max AE) and Maximum Absolute Percentage Error (Max APE) values. An improvement in the accuracy is observed using linear program (LP) compared to regression and artificial neural network (ANN). The flexibility of using different objective functions is observed in linear program model which is demonstrated by minimizing the maximum error.