An analysis of the 2021 real estate in Miami. Predicts accurate sale price on new data with a $50,000 margin of error on the test data. Many additional external factors need to be considered and integrated with the current model before any dependable predictions can be made. Real-time updates optimal, and further tuning for accuracy needed.
This project is intended as a real estate business insights exercise, as well as comparsion in performance of various machine learning regression models.The steps taken and models performance are broken down in this repo's .ipynb file.
Two visualizations in the notebook will not load on Github, and require running on Colab or a local Jupyter instance to view.
- 13,932 samples with 15 predictors to train model on
- Reserving 60% of dataset for training (appx. 8300 samples)
- The remaining 40% split evenly into validation and test sets to prevent data leakage and allow the model to generalize better (appx. 2800 each)