House Price Prediction in Tehran 🏠

This project uses linear regression to predict house prices in Tehran based on apartment features such as area, number of rooms, parking availability, warehouse availability, and address.

Dataset 📊

The dataset, sourced from "Apartments_information.csv", contains information about apartments in Tehran with the following columns:

🏠 Area: Apartment area in square meters (initially string, converted to float).
🛏️ Room: Number of rooms (integer).
🚗 Parking: Availability of parking (True/False).
📦 Warehouse: Availability of a warehouse (True/False).
🛗 Elevator: Availability of an elevator (True/False).
🏙️ Address: Apartment address (categorical, later encoded).
💲 Price(USD): Price in USD (float).

Initial size: 3479 rows; after preprocessing: 3452 rows.

Preprocessing ⚙️

-Loaded the dataset from "Apartments_information.csv".
-Dropped rows with missing values (23 rows dropped initially, then 4 more after Area conversion).
-Converted Area from string to float, replacing commas with dots.
-Assumed log transformations for Area and Price (as Area_log and Price_log) to handle skewness.
-Applied target encoding to Address within cross-validation, using the mean Price_log per address from training data.

Model 🧠

Algorithm: Linear Regression (sklearn.linear_model.LinearRegression)
Features:
🛏️ Room

📏 Area

🚗 Parking

📦 Warehouse

🏙️ Address_encoded

Target: Price

Evaluation: 5-fold cross-validation (KFold, shuffle=True, random_state=42)

Results 📈

R² Scores:

Fold 1: 0.854
Fold 2: 0.875
Fold 3: 0.837
Fold 4: 0.857
Fold 5: 0.856
Average R²: 0.856

MSE Scores:

Fold 1: 0.164
Fold 2: 0.142
Fold 3: 0.183
Fold 4: 0.172
Fold 5: 0.169
Average MSE: 0.166

Interpretation 💡

The model explains ~85.6% of the variance in Price_log (R² = 0.856), indicating a strong fit.
The MSE of 0.166 on the log scale suggests reasonable prediction accuracy.
A Pearson correlation of 0.81 between Area and Price highlights a strong positive relationship.

Usage 🚀

To run the code:

1.Ensure the required libraries are installed (numpy, pandas, matplotlib, seaborn, sklearn).
2.Place "Apartments_information.csv" in the working directory.
3.Execute the Jupyter notebook (House_price_prediction.ipynb).