Varsha Karunya Mekala

Toxic Comment Classification using NLP

• Developed deep learning model to classify social media comments into toxic categories like identity hate, obscene, threat, poison, and insult
• Achieved 99.47% accuracy and 91.2% AUC on dataset of 159,571 training comments and 63,987 testing comments.
• Developed detailed reports for management, providing insights into project progress and potential opportunities.

Suspicious Activity Report Detection in Financial Institutions using Machine Learning

• Developed classification model to predict SAR occurrences using customer data
• Implemented 41 distinct models with unique variable combinations using MATLAB's DOE tool.
• Conducted market research to identify industry trends and inform strategic planning initiatives.

• Conducted exploratory analysis of Pet Records and Donations data from Portage APL using Power BI
• Conducted data analysis to answer key questions posed by organization and identify trends and insights.
• Assisted management in making informed decisions by presenting clear and concise reports detailing analytical findings.

• Utilized Power BI to analyze customer data, resulting in 15% increase in customer retention rates
• Implemented Power BI dashboards, reducing report generation time by 30%.
• Improved business processes by analyzing current practices and recommending optimization strategies.
• Reduced operational costs through thorough data analysis and implementing cost-effective solutions.