Evaluation On Dhallywood Movies Based On Machine Learning

Abstract

The film is an exciting source of investment for passionate movie makers. The profitable nature of motion picture industry attracts movie creators to involve with it. In a such scenario, this is very important to evaluate movie status to find relevant features of a movie that make it successful. Machine learning is a popular trend for analyzing movie data. In our proposed research, we have tried to evaluate the status of Dhallywood movie based on three different class classifiers such as: Binary class (Hit-Yes, No), Triple class (Excellent, Good, Bad), and Four class (Excellent, Very Good, Good, Bad). The method of analyzing data has been described in details. Collection of Dhallywood movie data is the main challenge of this research work. The collected data have analyzed in a different way to set a target variable, which has improved the accuracy of models. The collected data have analyzed using five ML algorithms, and each algorithm applied three times for three different groups of class. Then the analytical results have compared to find out the best algorithm. From the comparative analysis it is found that accuracy of Triple class classification is higher than that of Binary and Four class classification. In addition to that among all applied algorithms, Random Forest provides the highest accuracy which is near about 85%. This research provides a new approach to set target variable classes based on Wikipedia data, news, actor actress biography, and viewer response on YouTube for a particular movie. We have selected this approach because the Dhallywood movie rating is not accurate on IMDb for all movies due to lack of budget and revenue data.