The objective of this project is to build a binary classification model for determining the validity of online claims submitted by customer for some product. The concept can be taken further to implement smart agents acting as filters to decide on the validity of customer requests coming in for replacements or returns and thus reducing the response time for a given case and improving customer satisfaction.
The Major Roadblocks faced:
- Class Imbalance
- Small Datasets for training
- Messy Text Data