Welcome to my portfolio of data projects. This repository includes a selection of hands-on projects using SQL and Python for real-world data analysis tasks. These projects cover exploratory analysis, data cleaning, and correlation analysis—fundamental skills in the data analytics and engineering domains.

Dataset

• Description:
  This project demonstrates exploratory data analysis using SQL on a structured dataset. It includes operations such as:

  • Handling and replacing NULL values, Standardizing formats (e.g., date and text case), Removing duplicates, Creating temporary clean views and applying transformation logic, Renaming columns for clarity and better readability
  • Identifying trends through grouping and filtering
  • Calculating statistical aggregates like averages, counts, and percentages
  • Using CASE statements to classify data
  • Applying conditional filtering to uncover insights

  The goal was to simulate a typical data exploration workflow within a relational database before building dashboards or models.

• Skills Practiced : Data cleaning, SQL scripting, formatting, NULL handling, view creation, aggregations, conditional logic, pattern recognition, CTE and window functions

• STACK: MySQL (PostgreSQL-compatible)

• Dataset

• REPORT

• Jupyter notebook with code

• Description:
  This Jupyter Notebook investigates how different variables (budget, gross revenue, runtime, etc.) relate to each other in a movie dataset. Steps included:

  • Data cleaning and preprocessing using pandas
  • Visualizing distributions and relationships with plotly.express
  • Calculating correlation and plotting a heatmap
  • Identifying which variables strongly affect a movie’s box office success

  The notebook helps answer business questions like: "Do bigger budgets lead to higher gross?" or "Which factors most influence revenue?"

• Skills Practiced: Exploratory data analysis, feature correlation, data visualization, Python scripting

• STACK: Python, Pandas, Plotly.express, Jupyter Notebook

• Description:
  This project evaluates the effectiveness of a new landing page through A/B testing. The analysis includes:

  • Comparing conversion rates between control (Group A) and treatment (Group B)
  • Conducting a z-test to assess statistical significance
  • Visualizing results and drawing business conclusions

  The goal was to simulate a real-world decision-making process using hypothesis testing.

• Skills Practiced: A/B testing, hypothesis testing, statistical analysis, data visualization

• STACK: Python, Pandas, SciPy, Seaborn, Matplotlib

• Dataset

• Jupyter notebook with code

• Description:
  This project analyzes the retention and purchase behavior of customers for an e-commerce company using cohort analysis. The notebook walks through:

  • Cleaning and transforming user transaction data with pandas
  • Assigning cohort groups based on user acquisition month
  • Calculating retention rates across cohorts
  • Visualizing cohort retention heatmaps with seaborn and matplotlib
  • Deriving insights about user engagement and drop-off patterns

  This type of analysis is commonly used by product teams and marketing analysts to understand customer lifecycle, evaluate loyalty, and improve retention strategies.

• Skills Practiced: Cohort analysis, retention analysis, data cleaning, data visualization, pandas transformations

• STACK: Python, Pandas, Matplotlib, Seaborn, Jupyter Notebook

• Description:
  This project automates the processing of raw CSV files to prepare them for analysis. It includes:

  • Reading input CSVs and handling encoding issues
  • Cleaning data by removing empty rows and standardizing column formats
  • Filtering and transforming specific fields (e.g., renaming, converting values)
  • Saving the cleaned output to a new CSV file

  The goal was to streamline repetitive preprocessing steps in data workflows.

• Skills Practiced: Data preprocessing, automation, file handling, basic ETL logic

• STACK: Python, Pandas, CSV module

• Description:
  Asynchronous API that recursively scrapes Wikipedia articles (up to 5 levels deep) and generates AI-powered summaries using DeepSeek API. Parsed articles, their relationships, and summaries are stored in a PostgreSQL database.

• Features:

  • Recursive Wikipedia parsing with parent-child relations
  • AI-generated summaries for articles
  • Async FastAPI backend with PostgreSQL
  • Fully Dockerized for easy deployment

• Skills Practiced:
  Async Python, API development, web scraping, AI API integration, PostgreSQL, Docker

• STACK:
  Python, FastAPI, SQLAlchemy (async), PostgreSQL, BeautifulSoup, DeepSeek API, Docker

• Description:
  End-to-end object detection pipeline for recognizing 6 classes of dishes and tableware in restaurant settings. Built using YOLOv11 with custom training, evaluation, and video visualization. Data preparation includes manual annotation, augmentation, and class balancing.

• Features:

  • Frame extraction from video (1 frame / 5 seconds)
  • Manual annotation with LabelImg (YOLO format)
  • 6 dish categories: tea, salads, kebab, chicken steak, soup, empty dishes
  • Data augmentation with Albumentations
  • Two-stage training with YOLOv11n and evaluation via mAP, Precision, Recall, F1
  • Output visualization in labeled video
  • Result analysis with per-class metrics and error matrix

• Skills Practiced:
  Object detection, dataset creation, model training & tuning, metrics analysis, OpenCV, Albumentations, YOLOv11

• STACK:
  Python, OpenCV, LabelImg, Albumentations, YOLOv11, PyTorch

⭐ If you find these projects helpful, please give the repository a star and feel free to connect!