Piyush Ratan Logo
    HomeWorkToolsAboutContact
    About MeView Tools
    All Work

    I implemented the models and compared evaluation metrics

    REGRESSION MODELS

    A machine learning foundations project comparing Linear Regression, Decision Tree, and Random Forest with proper evaluation instead of button-click ML.

    REGRESSION MODELS project visual

    Project Type

    Machine learning foundations project

    Stack

    Python, scikit-learn, NumPy, Pandas, Matplotlib

    Models

    Linear Regression, Decision Tree, Random Forest

    Timeline

    Built in 2026

    Case Study

    Engineering Notes

    01

    Project Overview

    Regression Models is a machine learning foundations project where I compare Linear Regression, Decision Tree, and Random Forest on regression tasks. The audience is mostly me as a learner, but the project also shows how I approach ML beyond surface-level imports.

    02

    Problem / Motivation

    It is easy to run scikit-learn examples and feel productive without understanding bias, variance, residuals, or why metrics disagree. I wanted a project that forced me to compare model behavior and not treat accuracy like the only truth in the room.

    03

    Architecture / System Design

    The workflow uses Python with NumPy, Pandas, scikit-learn, and Matplotlib. Data is loaded and cleaned, features are prepared, models are trained, predictions are generated, and evaluation metrics are compared. Visualizations help inspect prediction quality instead of relying only on terminal output.

    The structure stays intentionally simple: dataset preparation, model training, evaluation, and comparison. This keeps the focus on the ML fundamentals rather than hiding everything behind an oversized app shell.

    04

    Key Features

    The project is small, but it is built to make the fundamentals visible.

    • Linear Regression baseline implementation.
    • Decision Tree and Random Forest comparison.
    • MSE, MAE, and R2 evaluation.
    • Data inspection and visualization with Pandas and Matplotlib.
    • Notes around OLS, loss geometry, and gradient descent concepts.

    05

    Technical Challenges

    The challenge is not just running the models; it is interpreting them honestly. Different metrics can tell different stories, and tree-based models can look better while hiding overfitting risk. ML has enough traps. No need to add self-delusion also.

    06

    Solutions / Engineering Decisions

    I used a direct notebook/script style because this is a learning project. The decision was to keep the pipeline readable and metric-focused, then compare models through actual outputs instead of dressing the project up as a fake production system.

    07

    Outcome / Final State

    The project gives me a clearer base for regression workflows and model evaluation. It is not pretending to be a deployed ML product; it is a disciplined foundation for understanding the models I will later use in bigger systems.

    Machine LearningPythonscikit-learnNumPyPandasMatplotlib

    Key Capabilities

    Compared Linear Regression, Decision Tree, and Random Forest models.

    Evaluated models using MSE, MAE, and R2.

    Studied OLS derivation, loss geometry, and the path toward gradient descent.

    Links

    GitHub

    Seedha kaam. What I built, how it runs, what stack I used, where the control sits.

    More

    Keep Moving

    All Work

    2026

    PCM Syllabus Tracker

    A brutalist, high-performance spaced repetition tracker for Class 11 and 12 Physics, Chemistry, and Mathematics (PCM).

    2026

    DARPAN360

    A Chatbot-as-a-Service platform where users create domain-specific bots, lock allowed domains, and embed a controlled AI widget on real websites.

    2026

    NCERTGPT

    A RAG-based study assistant for Class 12 NCERT material, built around textbook extraction, chunking, embeddings, retrieval, and source-grounded answers.

    © 2026 Piyush Ratan. All Rights Reserved.