PRAMIT KUMAR BERA

A statistician by training with extensive exposure and in depth understanding of statistical modeling and classical machine learning methodologies. An innovative and knowledgeable professional having experience as Guest Lecturer/Faculty with proficiency in developing new lessons and activities to expand learning opportunities.

• Secured First Position/Gold medal in M.Sc. Statistics, Department of Statistics, West Bengal State University in the session 20102012.
• Secured Second Position in Poster Competition, Department of Statistics, National Level Statistical Seminar sponsored by University Grants Commission (UGC) & Ministry of Statistics and Program implementation, Govt. of India.

• Complete Guide to TensorFlow for Deep Learning with Python-Udemy
• Data Science for Business Leaders: ML Fundamentals- Udemy
• Python for Data Scienceand Machinelearning BootCamp- Udemy
• Certified Balanced Scorecard Professional (CBSCP).
• Artificial Intelligence for Business
• SAS programming Statistical Analyst Certification

Real Estate Market Analysis Using Advanced Regression Techniques

This project builds a predictive model for housing prices using variables such as area, bedrooms, and parking to analyze key factors driving real estate values. The approach involves data preprocessing, which includes normalizing features, removing outliers, and checking for multicollinearity using Variance Inflation Factor (VIF).

To improve prediction accuracy, multiple linear regression, Lasso, and Ridge regression models are employed, each offering unique advantages for capturing relationships between variables and addressing feature selection. Python libraries like pandas, scikit-learn, and statsmodels provide essential tools for these analyses. The model’s performance is evaluated through metrics such as R-squared, Mean Squared Error (MSE), and cross-validation.

The project aims to highlight factors most influential in determining property prices, providing practical insights for real estate agents, buyers, and investors. By leveraging advanced regression techniques, this work also demonstrates the potential of predictive analytics for applications in various industries.

Credit Card Fraud Detection
This project aimed to detect credit card fraud, tackling the problem that fraud cases are much rarer than normal transactions. To address this imbalance, it used SMOTE, a technique to increase the number of fraud cases in the data, which helps the model learn to identify fraud better.

Using Python libraries Scikit-learn and Pandas, the project compared two models: logistic regression and random forest. The random forest model proved better at detecting fraud, based on metrics like recall, which is essential for spotting actual fraud cases. This project shows that using the right methods and evaluation measures can make a big difference in detecting fraud accurately.

Title: EFFECTS OF YOGA ON DIFFERENT AGES 

Abstract: (Courses offered by a Wellness Center for 6 months) In this project we studied the effects of Yoga on persons of age group 41 to 60 years. For that purpose, we estimated the probability of persons who were satisfied by this course (Study variable) through the change of age (Auxiliary variable). But some people left the course before 6 months. For analyzing this issue, four models were chosen viz. Linear Probability Model, Logistic Regression Model, Likelihood Procedures and Censoring. Through this project, we were able to demonstrate that censoring is more efficient than the other three models. 

Supervisor: Dr. Shivnarayan Guria, Associate Professor, Dept. of Statistics, West Bengal State University.