ML1: Introduction to Machine Learning — House Pricing

Data Science Bootcamp is completed. Let me remind you my path:

• Data Science Bootcamp (completed)
• Core track (12 projects on key DS and ML topics)
• Practice (8 projects)

This project marks my official entry into the field of Machine Learning. While previous modules focused on general-purpose tools and data manipulation, this is where the real journey begins: transitioning from observing static charts to building algorithms capable of predicting future outcomes.

Topics

• Foundational ML: Understanding the conceptual shift from Rule-Based to Learning-Based systems.
• Supervised Learning: Practical implementation of Regression and Classification tasks.
• Exploratory Data Analysis (EDA): Identifying patterns, correlations, and anomalies in real-world data.
• Model Benchmarking: Establishing Baselines using naive models to measure true algorithmic value.
• Performance Metrics: Using MAE and RMSE to quantify and interpret prediction accuracy.

---

Roadmap

1. Foundations & Hypotheses

The work began with a theoretical deep dive. I explored the fundamental classification of tasks, determining why predicting rental prices is a Regression problem and defining the boundaries between multiclass and multilabel classification.

2. Statistical Data Auditing

One does not simply "feed" raw data into a model. I conducted a rigorous audit of the Kaggle RentHop dataset:

• Target Analysis: Visualized price distributions and identified extreme outliers that could compromise model integrity.
• Data Cleaning: Applied statistical filtering (1st–99th percentile) to strip away noise and focus on the representative data range.
• Correlation Study: Utilized Heatmaps and Scatterplots to quantify how features like bedrooms or bathrooms actually impact the final cost.

3. Feature Generation

I experimented with data complexity by creating Polynomial Features up to the 10th degree. This stage was essential for understanding the trade-off between model flexibility and computational cost, observing how feature transformations can capture non-linearities.

4. The Model Showdown

To establish a performance baseline, I trained and compared three distinct approaches:

• Linear Regression: The primary baseline for predicting continuous values.
• Decision Tree Regressor: An introduction to non-linear, tree-based data splitting.
• Naive Models (Mean/Median): A critical "sanity check" to ensure that the developed models provide genuine predictive power over simple averages.

---

Results

The project concluded with a comprehensive evaluation using MAE and RMSE. By comparing error rates across different architectures, I identified the most effective model for this dataset. This module taught me that Machine Learning is not just about writing code; it’s about asking the right questions and methodically searching for answers within the data.

How to Run the Project

1. Clone the repository:

   git clone https://github.com/knight99rus/ML1_Introduction.git
   cd ML1_Introduction

2. Create and activate a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # For Windows: venv\Scripts\activate

3. Install dependencies:

   pip install jupyter pandas numpy scikit-learn matplotlib seaborn scipy statsmodels lightgbm

4. Download data:

   • Read the task on the Kaggle competition page.
   • Download test.json file.

5. Launch Jupyter Notebook:

   jupyter notebook

   Open and execute the cells in the project01.ipynb file.