Sohini Das

Optimize fresh produce wastage focusing on fruits and vegetables. Achieved a predictive accuracy rate of 86.59% using Random Forest and Decision Trees. Implemented multi-layered classification approaches using SMOTE, social network analysis, and XGBoost. Integrated insights into decision-making processes to support strategic planning.

Collected and preprocessed 5 years of purchase history data from Azure Data Lake and SQL server. Developed and fine-tuned machine learning models using statistical tests and K-Means clustering. Transformed data into PowerBI reports using Databricks and Pyspark. Maintained Data Factory pipelines, leading to a 25% reduction in infrastructural cost.

Executed web scraping on e-commerce platforms for competitive benchmarking. Cleaned and pre-processed data for a Predictive Regression Model. Automated MySQL-Python pipeline, reducing infrastructure costs by 65%.

Authored Data Science articles covering foundational Hypothesis Testing to advanced Convolutional Neural Networks (CNN). Initiated a peer review culture and co-authored ventures across blog platforms.

Assisted in predictive sales reporting and Data Quality Assessment. Analyzed US vehicle purchase data, achieving 93% accurate purchase tendency forecast.

Led a team of 3 in the development of a bid forecasting and sales prediction system. Designed a Python-Sklearn-MySql-Tableau integration framework for visualization and prediction.