SREEJAN SAHA

Range prediction for quality controlling parameters of a tablet. Predicted tablet thickness with Random Forest Regression Thickness model having 77% accuracy along with variable explain ability. Predicted tablet hardness with Random Forest Regression Hardness model having 72% accuracy along with variable explain ability. Perceived value ranges of parameters important for quality tablet production. Detection of successful Promotional Pattern for newly launched Medicine. Promotional patterns of each successful and failure drug is identified using Pattern Mining. Developed Python scripts and implemented Pattern-growth-based-approach Prefixspan to understand the sequence patterns of promotional events that caused success. Developed Python scripts and used frequency pattern mining approach Pyfp Growth to identify most frequent market promotional events.

Identification of Restricted Names in Bank data. Structured the data in data frame with two columns. Used TF-IDF method to prepare labelled data, matched string are marked as 1, and rest as 0. Used Feature engineering for feature extraction(e.g. Removed stop words, count of common words, total string length). Used SMOTE to resample data and applied Ensemble techniques in process and achieved highest 72% accuracy.