🌐 Data Scientist | 👷🏻♂️ Ex. Senior Engineer (Mech.) | 🌍 From Agartala, Tripura
- 🔍 Exploring the world of Data Science, Machine Learning, and Business Analytics
- 💼 Ex-Senior Engineer(Mech) at NEGG Project ( PM GATI SHAKTI )
- 🎓 B.Tech in Mechanical Engineering
- 💡 5⭐ in SQL on HackerRank | Real-World Impact | EDA Enthusiast
- 📈 Passionate about transforming data into meaningful insights
-
🔍 Objective: To analyze and detect fraudulent financial transactions using machine learning, with a focus on real-time prediction, interactive visualization, and explainability.
-
🛠️ Tools: Python, Scikit-learn, Streamlit, Pandas, Seaborn, Joblib
-
🔄 Process:
-
Loaded and cleaned financial transaction data from kaggle; checked missing values, types, and class imbalance.
-
Engineered features such as balance differences and flagged suspicious patterns like zero balances post-transfer.
-
Visualized fraud distribution by transaction type and time using Seaborn and Matplotlib.
-
Built a classification pipeline with Logistic Regression (class_weight='balanced'), and evaluated using confusion matrix and classification report.
-
Saved trained model with Joblib for deployment.
-
-
💡 Insights:
-
Fraud was highly concentrated in specific transaction types with sharp balance changes.
-
Feature engineering improved model precision and interpretability.
-
Visual exploration helped uncover hidden patterns related to fraud triggers.
-
-
✅ Results:
-
Achieved ~94% accuracy with strong precision-recall balance for imbalanced fraud detection.
-
Deployed a Streamlit web app for real-time fraud prediction and interactive data exploration.
-
App allows users to upload new transaction data and visualize predictions instantly.
-
- 🔍 Objective: To build a regression model to accurately predict Mexico housing prices using location, surface area, and property type features.
- 🛠️ Tools: Python, PowerBI
- 🔄 Process :
- Collected real estate data including price, location, area, and property type from Mexican housing listings.
- Cleaned data by removing nulls, duplicates, and extreme outliers; engineered features like price_per_sqm.
- Performed EDA using plots and correlation heatmaps to uncover key variables affecting price.
- 💡 Insights :
- Housing prices were significantly higher in capital cities and tourist areas.
- Surface area was positively correlated with price, but gains plateaued after a certain point.
- Apartments had a higher price per square meter compared to standalone houses.
- ✅ Results :
- Ridge Regression reduced prediction error with MAE ~15,200 and better generalization than Linear Regression.
- Prepared dashboard-ready data in Power BI.
- 🔍 Objective: To explore job trends and skills demand in the data industry.
- 🛠️ Tools: Python, SQL and Excel.
- 📊 Process:
- Tried to collect job listings data using web scraping and APIs.
- Cleaned and normalized job titles, locations, and skill tags.
- Connected MySQL using SQL Connector for Queries.
- 💡 Insights:
- Management, Engineering and Analyst were top 3 demanded skills.
- Seoul and Apia were major hubs for Multiple Job Post.
- Most Jobs were posted in December-2021.
- Companies favored candidates with real-world project exposure.
- ✅ Result: Provided strategic recommendations for learners and job seekers.
- Deepening knowledge in Machine Learning and Deep Learning
- Exploring Natural Language Processing (NLP) and Prompt Engineering
- Building Real-World Data Science Projects
- Practicing Advanced SQL and Statistical Techniques
- Improving Dashboarding with Power BI
If you found my work interesting or useful, feel free to connect or reach out — I'm always open to learning and collaboration!
“The goal is to turn data into information, and information into insight.”