Home
AI/ML Projects
Machine Learning Project

Case Eligibility Analyzer Two-Stage Model

CEA is a two-stage machine learning model built to help law firms assess legal case eligibility and predict potential claim amounts. Designed to improve intake screening and resource allocation, it classifies whether a case is viable and estimates the likely settlement amount.
Share this post

Motivation

Law firms waste hours reviewing ineligible leads. I built this project to explore how AI could speed up case screening and help surface high-value cases faster.

Hugging Face Interface for Plant type model

Key Highlights

  • Built a two-stage pipeline:
    • Stage 1: Claim eligibility classification (82.3% accuracy)
    • Stage 2: Claim amount prediction (RMSE $4,075)
    Used Fastai tabular neural networks with dynamic undersampling and log transformationEngineered features from a large insurance claims dataset (67K+ rows)Deployed model architecture and UI prototype for demonstration on Hugging Face
Share this post
Design
Product
Research