Developed a machine learning model to predict the likelihood of project approval on the DonorsChoose platform. This model assists teachers in understanding how likely their projects are to receive financial support, thereby enabling them to optimize their project proposals for better outcomes.
- Data Source: Utilized a comprehensive dataset from Kaggle, which can be found here: DonorsChoose.org Application Screening Dataset.
- Prediction Goal: The model predicts the approval status of a project proposal, helping to identify the factors that increase the likelihood of funding.
- Model Performance: Achieved an Area Under the Curve (AUC) score of 0.7 using Bag of Words text featurization and 0.61 with Term Frequency-Inverse Document Frequency (TfIdf) methods, indicating a robust predictive capability.
- The model's performance metrics suggest that Bag of Words is a more effective method for text featurization in this context compared to TfIdf.
- Insights derived from the model's predictions can be used to guide teachers in crafting proposals that align better with funding criteria, ultimately increasing their chances of approval.
- By providing a predictive understanding of project approval chances, this tool can be a valuable asset for educators seeking funding on DonorsChoose.org.
- The model's insights have the potential to contribute positively to the educational community by supporting more effective and targeted project proposals.