Mangesh Patil

Proficient in Python, SQL, Tableau, and machine learning.

About me

I’m Mangesh Patil, a data scientist with expertise in analytics, machine learning, and predictive modeling. I specialize in building scalable data solutions to drive strategic decision-making and solve complex problems in areas like credit risk, customer insights, and fraud detection. My experience spans roles where I’ve developed credit risk models with 85% accuracy, optimized customer engagement through data-driven strategies, and built hybrid recommendation systems for personalized experiences. Proficient in SQL, Python, and cloud platforms, I thrive at the intersection of data and business, delivering measurable outcomes. Let’s connect to discuss how I can bring value to your data challenges.

Exploring Data Engineering with Microsoft Azure: Tokyo Olympics 2020 Project

  • This project was an invaluable learning experience, demonstrating the power and versatility of Microsoft Azure in handling various aspects of data engineering, from data collection and storage to processing and visualization.
  • By utilizing Azure Data Factory, Data Lake Gen 2, Databricks, and Power BI, I was able to build a robust pipeline that provided actionable insights from raw data.
  • Through this project, I gained a comprehensive understanding of how different Azure services can be integrated to create efficient data engineering solutions.
  • I am excited to apply these learnings to future projects and continue exploring the possibilities within the field of data engineering.

Dashboards [Power BI and Tableau]

  • Call Center Solution.pbix: Dashboard focusing on call center operations and metrics.
  • Customer Retention.pbix: Dashboard designed to analyze customer retention strategies and metrics.
  • Diversity & Inclusion.pbix: Dashboard presenting data and insights related to diversity and inclusion initiatives.
  • Movies Dashboard.pbix: Dashboard offering analysis and visualization of movie-related data.
  • Grocery Store Management and KPI.pbix: Dashboard providing insights and key performance indicators for a grocery store management system.

CreditRisk Modelling

  • Developed and implemented machine learning models, including logistic regression and XGBoost, to predict loan defaults, with a focus on improving accuracy and risk management.
  • Conducted comprehensive data analysis and preprocessing techniques, including handling missing data and performing feature engineering, which resulted in optimized model performance.
  • Evaluated model efficacy using key metrics such as accuracy, precision, recall, and F1-score, achieving robust predictive capabilities for loan default prediction.
  • Provided actionable insights into credit risk factors through feature importance analysis, thereby enhancing decision-making processes for financial institutions.

SkyData Insights: Flight Booking Analysis and Predictive Modeling

  • Spearheaded complex data cleaning, encoding, and feature engineering to generate high-quality datasets.
  • Constructed Random Forest and XGBoost classifiers, and evaluated models using classification reports and confusion matrices.
  • Applied techniques such as SMOTE oversampling to address class imbalance, significantly enhancing model performance.
  • Leveraged proficiency in Pandas, Scikit-learn, and XGBoost for advanced data manipulation and modeling, enabling sophisticated analysis.

Netflix Recomendation system

  • Led the development of a sophisticated Netflix recommendation system, incorporating advanced content-based, collaborative filtering, and hybrid methodologies.
  • Conducted intricate content-based analysis using techniques like TF-IDF and word embeddings on a dataset of over 10,000 movie titles and 25 million entries, contributing to the system's ability to personalize recommendations.
  • Integrated collaborative filtering algorithms, including User-Based and Item-Based Collaborative Filtering, achieving performance metrics such as an average RMSE of 0.86 and MAE of 0.70, enhancing recommendation accuracy and relevance.
  • Engineered a hybrid recommendation system, optimizing recommendation accuracy through techniques such as weighted averaging and ensemble methods, ensuring a seamless blend of different recommendation approaches.

Housing Value Navigator: Predictive Property

  • Conducted exhaustive analysis of California state housing data, leveraging tools like Python with pandas and NumPy for data manipulation.
  • Engineered and optimized machine learning models using TensorFlow for housing price prediction.
  • Employed regression models and ensemble methods for robust predictive capabilities.
  • Achieved a 15% enhancement in decision-making accuracy for real estate professionals, utilizing advanced analytics and model interpretation techniques.

Bank of America Job Simulation on Forage

  • Identified an ideal acquisition target for a client based on a SWOT analysis and assessment of their strategic criteria.
  • Constructed a DCF model to calculate the implied equity and share value of the acquisition target.
  • Completed a sensitivity analysis to illustrate how the target’s valuation would change as variables change.
  • Created a company profile summarizing all key information about the target.

Cognizant's Artificial Intelligence Job Simulation on Forage

  • Completed a job simulation focused on AI for Cognizant’s Data Science team.
  • Conducted exploratory data analysis using Python and Google Colab for one of Cognizant’s technology-led clients, Gala Groceries.
  • Prepared a Python module that contains code to train a model and output the performance metrics for the Machine Learning engineering team.
  • Communicated findings and analysis in the form of a PowerPoint slide to present the results back to the business.