House Price Prediction

Overview

This project predicts house prices based on key features such as size, location, and age. It leverages advanced machine learning techniques, including XGBoost, to deliver accurate price estimates. These insights can assist real estate agents, buyers, and sellers in making informed decisions.

Key Features

• Predictive Model: Built using XGBoost, optimized for tabular data.
• Feature Engineering: Created custom features like house age and area ratios to boost model performance.
• Performance Metrics: Achieved competitive MAE and RMSE scores.
• End-to-End Workflow: Includes data preprocessing, exploratory analysis, modeling, and deployment.

---

Results

• Mean Absolute Error (MAE): 15,734
• Root Mean Squared Error (RMSE): 125

The model provides reliable price predictions, reducing uncertainty in property valuation.

---

Project Structure

House_price_prediction/
├── house/                  # Dataset folder
├── notebooks/              # Jupyter notebooks
│   ├── Visual.ipynb        # Exploratory Data Analysis
│   └── XGB_regress.ipynb   # Model training and evaluation
├── models/                 # Saved models
│   ├── xgb_model.pkl       # Main XGBoost model
│   └── trained_model.pkl   # Alternative trained model
├── scripts/                # Utility scripts
│   ├── save_model.py       # Script for saving models
│   └── deploy_model.py     # Script for deployment
├── requirements.txt        # Dependencies
└── README.md               # Project overview

---

Data

The dataset contains key features influencing house prices:

• Size: Total area in square footage.
• Bedrooms & Bathrooms: Count of each.
• Location: Neighborhood information.
• Year Built: Construction year.
• Sale Price: Target variable for prediction.

Data Preprocessing:

• Imputation for missing values.
• Encoding for categorical variables.
• Scaling for numerical features.

---

Modeling

Model: XGBoost

• Chosen for its speed and performance on tabular datasets.
• Trained using advanced hyperparameters:
  • Learning Rate: 0.05
  • Max Depth: 4
  • Subsample: 0.8
  • Colsample by Tree: 0.2
  • n_estimators: 200

Validation

• Applied cross-validation to ensure consistent results.

---

How to Run

Prerequisites

Install required libraries:

pip install pandas numpy scikit-learn xgboost matplotlib seaborn

Installation

1. Clone the repository:

   git clone https://github.com/NasdormML/House_price_try.git
   cd House_price_try

2. Install dependencies:

   pip install -r requirements.txt

Usage

• EDA: Run Visual.ipynb to explore the dataset and trends.
• Model Training: Use XGB_regress.ipynb to train and evaluate the predictive model.

---

Business Value

• Accurate Pricing: Helps set realistic property prices, increasing transaction efficiency.
• Market Insights: Identifies key factors driving property value.
• Risk Reduction: Assists buyers in avoiding overpayment.

---

Contact

Have questions or feedback? Reach out via email: nasdorm.ml@inbox.ru

---