PROJECTS

Telemarketing Campaign Response Prediction

Link to Project Repository:
https://github.com/armanic12/Telemarketing_Prediction_Model_Testing

Summary

This project applies machine learning to predict customer responses in a telemarketing campaign using the UCI Bank Marketing dataset. Data preprocessing included encoding, normalization, and SMOTE balancing. Multiple models (KNN, Decision Tree, SVM, and Neural Network) were evaluated, with the hypertuned neural network achieving the highest accuracy and AUC.

Skills & Tools Applied:

• Data handling and preprocessing with pandas and NumPy

• One-hot encoding and normalization with scikit-learn

• Class balancing using SMOTE (Synthetic Minority Over-sampling Technique)

• Model development with K-Nearest Neighbors, Decision Tree, and Support Vector Machine

• Hyperparameter optimization using GridSearchCV and Keras Tuner Hyperband

• Neural network implementation with TensorFlow/Keras (Sequential API)

• Model evaluation with accuracy, precision, recall, F1-score, ROC curve, and AUC

• Data visualization with Matplotlib and Seaborn

Amazon Product Recommendation Modeling

Link to Project Repository:
https://github.com/armanic12/Amazon_Product_Recommendation_Modeling/tree/main

Summary

This project implements an item-based collaborative filtering recommendation system to suggest Amazon Beauty products based on a user's past ratings and similarity to other users' preferences. Using the Amazon Beauty product ratings dataset, the workflow includes loading and cleaning the data, constructing a user-item matrix, applying dimensionality reduction via TruncatedSVD, and computing product correlations to identify and return closely related products.

Skills & Tools Applied:

• Data ingestion and preprocessing with Python, pandas, and NumPy

• Construction of item–user utility matrix for collaborative filtering

• Dimensionality reduction using scikit-learn’s TruncatedSVD

• Correlation analysis to compute item–item similarity with NumPy

• Implementation of item-item collaborative filtering logic

• Basic data visualization for insight validation with Matplotlib

NBA 2024-25 Analysis

Link to Project Repository:
https://github.com/armanic12/2024_25_NBA_Analysis_Project

Summary
This project extracts, cleans, and exports NBA player per-game data for the 2024-2025 season. It feeds the cleaned data into a Tableau dashboard for exploratory analysis and interactive visualization.

Skills & Tools Applied:

• Web scraping with Python

• HTML parsing using BeautifulSoup

• Data extraction from structured HTML tables

• Data cleaning with regex and string operations

• Data transformation and organization using pandas

• API-based file export to GitHub

• CSV file handling and local export workflows

• Data visualization design in Tableau

• Integration of scraped datasets into interactive dashboards