Using the Ames, Iowa Housing Dataset I build a regularized regression model with Elastic Net.
This project has two objectives
-
Build and train an elastic-net-regularized regression model to predict house prices
-
Develop a standardized framework to structure the implementation of this model and models in future machine learning projects
In the process of developing a standardized framework, most of my time has been spent honing my skills with Scikit Learn pipelines. Pipelines provide a common framework to structure all steps of a data science project, from data preprocessing to prediction. By adhering to a standard framework more time may be spent on data analysis, feature extraction, and understanding the statistical models, and less time spent implementing these things.
This project depends on thermidor. thermidor is a Python module containing several functions and classes which simplify the creation of machine learning projects by streamlining Sci-kit Learn pipeline construction.
.
├── 1.1-enet-house-prices.ipynb
├── LICENSE
├── README.md
├── data
│ ├── data.csv
│ └── response.csv
├── models
│ └── enet_model.pkl
└── predictions
└── predictions.csv
The project itself may be found at ./1.1-enet-house-prices.ipynb. The trained and fitted model may be found at ./models/enet_model.pkl.
Created by Richard Correro in 2019. Contact me at rcorrero at stanford dot edu