MichelleWambaya/REAL_ESTATE-Analysis

This project presents a data analytics case study using a fictional real estate dataset from Kenya. It aims to extract insights and build predictive models that estimate property prices using machine learning. # Real Estate Data Analytics in Kenya – Project Notes This project presents a data analytics case study using a fictional real estate dataset from Kenya. It aims to extract insights and build predictive models that estimate property prices using machine learning. --- ## 1. Project Overview This notebook simulates a dataset for Kenyan housing properties and applies analytics to: - Understand housing market trends - Identify key features affecting pricing - Predict property prices using machine learning --- ## 2. Data Generation A synthetic dataset was generated to resemble real-world Kenyan housing data, including: - `Bedrooms`, `Bathrooms`, `Size_sqm`, `Proximity_CBD_km` - `Location`: Nairobi, Mombasa, Kisumu, Eldoret - `House_Type`: Apartment, Bungalow, Maisonette - `Price_KES`: Simulated price values based on features The data was generated with logical patterns, such as larger or better-located houses costing more. --- ## 3. Exploratory Data Analysis (EDA) Key insights from the dataset: - **Size and price** are positively correlated - **Proximity to CBD** inversely affects price - **Nairobi and Mombasa** tend to have higher average prices - **House type** influences pricing (e.g., Maisonettes are more expensive) Tools used include bar plots, scatter plots, boxplots, and correlation heatmaps. --- ## 4. Feature Engineering Selected features: - `Bedrooms`, `Bathrooms`, `Size_sqm`, `Proximity_CBD_km` - Target: `Price_KES` These features are both interpretable and relevant to home buyers and sellers in Kenya. --- ## 5. Machine Learning Models ### 🔹 Linear Regression - Assumes a linear relationship between features and price - Easy to interpret and a good baseline - Metrics: - **Mean Absolute Error (MAE)** - **R² Score** ### 🔹 Random Forest Regressor - Ensemble model combining multiple decision trees - Captures non-linear relationship …