Vaibhav Avhad

Project 1 - Customer CHARN Analysis

• Analyzed 5 years of customer behavior data to find patterns in recovery rates, identify top-performing regions and determine the best ways to connect with customers—turning insights into action.
• Segmented customers based on historical repayment behavior, delays, spending patterns, and customer connectivity to identify high-risk profiles.
• Analyzed geographic trends to determine regions with higher recovery rates and compared their performance with underperforming areas.
• Evaluated the impact of different recovery actions (calling, field visits, legal actions) on repayment success.
• Optimized monthly resource allocation using data-driven insights to improve recovery efficiency and enhance decision-making for delinquent accounts.
• Improved repayment rates by 1.5 to 5.5%, adding ₹70-80 Lakh per month to company revenue by identifying high-risk customers for targeted recovery strategies, enhancing credit risk management.
• Project 2 - Loan Default Prediction Model
• Tool & technique: Excel, SQL, AWS, Python(Pandas, NumPy, SciPy, Scikit-learn, Matplotlib, Seaborn).
• Developed a Decision Tree model achieving 85% accuracy in predicting loan defaults.
• Provided actionable insights into credit risk management, aiding resource allocation for high-risk customer recovery.

• Utilized tools and technologies such as Python (NumPy, Pandas, Sklearn, SciPy), Matplotlib, Seaborn, Excel, and PowerPoint for data collection, Data handling, EDA,and visualization.
• Collected and analyzed time-series data from over 1,00,000 samples, using advanced techniques to identify trends and provide actionable insights in detailed reports.
• Collaborated as part of the animal experiment team, analyzing and interpreting data from laboratory experiments to ensure accuracy and reliability for further research applications.
• Designed user-friendly dashboards to elegantly visualize data insights, increasing accessibility for cross-functional teams by 30%.

Project: Health Recommendations Model

• Built an ML model to predict user health conditions and provide personalized recommendations for diet, fitness tips, meditation, music, and yoga.
• Implemented Random Forest achieving 85% accuracy in predicting health conditions based on user input data. Enabled users to receive data-driven wellness suggestions tailored to their needs.

• Statistical Analysis, Data Visualization, Hypothesis testing.
• Dashboard preparation in Tableau Desktop.

Project

• House price prediction.
• Diamond price prediction.
• Image prediction using Keras.
• Time Series forecasting with LSTMJune