Kickstarter, a renowned crowdfunding platform, operates on a unique premise where backers support projects of interest through financial pledges. The platform employs an ”all or nothing” model, where each project sets a financial goal, and its outcome is categorized as either failed or successful based on whether the goal is achieved. The high stakes associated with this model underpin the importance of a predictive model that can accurately forecast a project’s fate. Such a tool would be invaluable for project creators, allowing them to assess the suitability of Kickstarter for their endeavor before committing, ultimately saving time and resources. Moreover, delving into the diverse attributes of past projects can provide creators with insights to strategically position their initiatives for success. Recognizing this potential, this project aims to develop a classification model capable of predicting a project’s success or failure. Additionally, the project seeks to employ clustering techniques on historical data to uncover inherent patterns and trends among past Kickstarter projects, offering creators a deeper understanding to enhance their project planning and execution strategies.
- Python
- Pandas
- Matplotlib
- Seaborn
- Boosting and Bagging
- K-Prototypes clustering
- SHAP