Project 3

Computer Price Prediction

Machine Learning[ Apr 2026 - May 2026 ]

Overview

Engineered a predictive model using a Random Forest Regressor to estimate custom computer prices based on complex hardware specifications, following the comprehensive CRISP-DM methodology across 100,000+ dataset rows with rigorous feature selection.

Key Features

  • End-to-end CRISP-DM pipeline structuring data understanding, preparation, modeling, and evaluation phases in Python
  • IQR Winsorization for outlier treatment preserving data stability and distribution integrity
  • Wrapper (RFE) feature selection method minimizing multicollinearity and optimizing dimensionality
  • 87.85% R-squared test accuracy and $157.13 Mean Absolute Error, outperforming KNN and SVR baselines
  • 20 high-quality statistical visualizations using Matplotlib and Seaborn for data variance and feature importance mapping

Skills

Machine LearningPythonData AnalyticsPredictive ModelingData Science

Gallery

Project Info

CategoryMachine Learning
PeriodApr 2026 - May 2026
TagPython / CRISP-DM

Tech Stack

Machine LearningPythonData AnalyticsPredictive ModelingData Science