A prediction model using a decision tree to determine what an optimal price might be for a house, based on historic housing data from Boston. This project demonstrates the efficacy of using GridSearch algorithm to find optimal parameters for a learning model.
This project requires Python 2.7 and the following Python libraries installed:
In addition, you will need to be able to run an jupyter Notebook to go along with this project.
In a terminal/command window, go to the top-level project directory boston_housing/ (that contains this README). Then run:
jupyter notebook boston_housing.ipynb
The dataset used in this project is included with the scikit-learn library (sklearn.datasets.load_boston). You do not have to download it separately.
It contains the following attributes for each housing area, including median value (which you will try to predict):
- CRIM: per capita crime rate by town
- ZN: proportion of residential land zoned for lots over 25,000 sq.ft.
- INDUS: proportion of non-retail business acres per town
- CHAS: Charles River dummy variable (= 1 if tract bounds river; 0 otherwise)
- NOX: nitric oxides concentration (parts per 10 million)
- RM: average number of rooms per dwelling
- AGE: proportion of owner-occupied units built prior to 1940
- DIS: weighted distances to five Boston employment centres
- RAD: index of accessibility to radial highways
- TAX: full-value property-tax rate per $10,000
- PTRATIO: pupil-teacher ratio by town
- B: 1000(Bk - 0.63)^2 where Bk is the proportion of blacks by town
- LSTAT: % lower status of the population
- MEDV: Median value of owner-occupied homes in $1000's