Sandeep Mane

Experience

1. Data Analytics

   Kotak Mahindra Bank

   Dec 2021 — Present · Thane, IN

   Credit Card (CC) Risk:• Developed a Credit Card (CC) scorecard for a partner’s customer base, the goal is to predict eligibility based on a combination of policy filters and credit risk modeling.• Achieved 89.1% training accuracy and 85.4% validation accuracy.• Optimized model performance by fine-tuning hyperparameters using RandomizedSearchCV across four iterations, enhancing predictive accuracy.• Surpassed baseline models with an AUC over 94% and 87% precision, demonstrating exceptional capability in identifying defaults.• Techniques used: Logistic Regression CD NTB (New-to-Bank) Customers (bureau hit):• Developed and monitored for application scorecard for CD NTB bureau hit customer which will help bank to identifying risky customer before giving them sanction for asked loans.• Achieved 87.6% training accuracy and 82.3% validation accuracy.• Optimized model performance by fine-tuning hyperparameters using RandomizedSearchCV across four iterations, enhancing predictive accuracy.• Surpassed baseline models with an AUC over 90% and 85% precision, demonstrating exceptional capability in identifying defaults.• Techniques used: Logistic regression and Random Forest. POS Machine Forecast:• Developed ARIMA and SARIMA models to forecast monthly sales trends with high accuracy, reducing forecast error by 17%.• Enhanced data quality and model reliability by implementing thorough data preprocessing and stationary transformations (differencing), achieving a 13% increase in improvement.• Technique : Time Series

Education

• Sardar Patel University

  M.Sc Statistics

• S.P.university Gujarat

  M. Sc

  2018 — 2020