The main idea is:
"Apply Machine Learn to predict Boston houses prices"
In this project used machine learning to predict Boston property prices, and to show the data was build a data app, able to interact and receive user information and predict the property prices from the input data.
Data app can be find here. (All informations are portuguese language 🇧🇷)
This dataset is about Boston houses prices, and can be get through sklearn library, you can get more information here
Attribute Information:
- 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
To predict a estimated value, i start a simple baseline, and apply Machine Learning algorithms like Linear Regression, Decision Tree and Random Forest to get better results.
The baseline metric is very simple (shown in code). Therefore so his prediction isn't so good, below the figure show a comparison between the Real Value and Baseline Prediction
For bests results, i apply Machine Learn algorithms, in order to get close of Real Value function.
To measure the real performance of the algorithms, the mean squared error is calculated.
- Baseline error:
6.21 - Linear Regression error:
4.46 - Dicision Tree error:
4.42 - Random Forest error:
3.33
The Random Forest algorithms was the best results along others. Then he will be the algorithm used to build the data app.