The objective of our project is to develop a predictive model that can forecast the popularity of songs on Spotify. To achieve this, we will leverage a comprehensive dataset containing various features related to each song. These features encompass details such as song attributes, information, and specific metrics from the Hot 100 chart.
- EDA: The project involved intensive EDA and data analysis to understand the dataset thoroughly.
- Preprocessing: Extensive preprocessing ensured data integrity by addressing nulls and not real data.
- Feature Engineering: Various feature engineering techniques were employed, including encoding, scaling, transformations, outlier handling, and the creation of new features.
- Feature Selection: Applied feature engineering using ANOVA, Pearson, and Spearman correlation methods.
- Model Training: Model selection was conducted utilizing automated hyperparameter tuning to enhance performance.
- Model Evaluation: A thorough evaluation process led to the identification of the best model.
Deployed via Streamlit, users can effortlessly predict song popularity using this link: https://the-song-popularity-predictor.streamlit.app/