Abstract: The music industry attracts a large number of investors due to its high turnover. Releasing hit songs can garner profits, whereas flop songs lead to losses. Thus, predicting the popularity of a song before its release can help in promotion plans. Can we really predict hit Songs? This is the main motivation of the work. Extensive work is being done for western songs, but Indian music is relatively less explored. Hence, our work aims to predict hit songs of Indian origin using acoustic features. To analyze tracks, a data set is created from data provided by the Spotify Web API. The features are extracted using Spotify, available libraries such as Librosa and aubio, which are passed to machine learning algorithms for prediction. Along with available features, melodic features based on patterns are proposed and extracted. A comparative analysis is done for four acoustic feature sets containing timbral, pitch, rhythm and melodic features proposed. Further experimentation is performed with the combined features sets resulting in the improved performance. The results are encouraging and hit song prediction can be a reality in the near future.