This project focused on developing predictive models to understand the factors influencing property values in Barcelona. The goal is to forecast future prices, maximizing returns on real estate investments and facilitating informed market decisions.
- Natalia Beltrán
- Guillem Mirabent Rubinat
- Ed Monbiot
The project aims to create three models—Linear, Lasso, and Ridge—to analyze historical data, market trends, and key determinants that impact property values. The primary objective is to develop a robust predictive model capable of accurately estimating apartment prices, thereby aiding in crucial investment decisions.
- Goals & Criteria: Establishing the project objectives and success metrics.
- Loading Libraries & Data: Setting up the necessary libraries and importing data for analysis.
- Data Exploration: Investigating dataset characteristics and underlying patterns.
- Identifying Anomalies & Missing Data: Handling data integrity issues to ensure the quality of analysis.
- Data Pre-processing: Cleaning and transforming data for effective modeling.
- Linear Model: Implementing a linear regression model for initial price predictions.
- Hyperparameter Tuning & Cross-Validation: Optimizing model parameters for enhanced predictive performance.
- Lasso Model: Applying Lasso regression for feature selection and enhanced prediction accuracy.
- Ridge Model: Utilizing Ridge regression to investigate its impact on prediction results.
- Conclusion: Summarizing insights and outcomes from the analysis.
- Python for comprehensive data analysis and modeling.
- Libraries including Pandas, NumPy, Scikit-learn for data manipulation, statistical modeling, and machine learning.
- Advanced techniques in cross-validation and hyperparameter optimization.
- The best model for predicting housing prices in Barcelona is our Ridge model with an MSE of 29165.372 and an alpha 0
- Top of the class: Team Sarria
- https://www.kaggle.com/competitions/predicting-apartment-prices-in-barcelona-2023/leaderboard