The predict_real_estate_prices from fannie-wong

Predicting Amsterdam house / real estate prices using Linear Regression, KNN-, Lasso-, Ridge-, Polynomial-, Support Vector (SVR)-, Decision Tree-, Random Forest-, and Neural Network MLP Regression.

Approach:

load Pandas DataFrame containing (Dec-17) housing data retrieved by means of the following scraper, supplemented with longitude and latitude coordinates mapped to zip code (via GeoPy)
do some simple data exploration / visualisation
remove non-numeric data, NaNs, outliers and normalise data
define explanatory variables (surface, rooms, latitude, longitude) and independent variable (price EUR)
split the data in train and test set for later usage
find the optimal model parameters using scikit-learn's GridSearchCV
fit the model using GridSearchCV's optimal parameters
evaluate estimator performance by means of 10 fold 'shuffled' cross-validation

Packages required

Results along (Dec-17) Amsterdam house / real estate price data retrieved by means of the following scraper

Sample data input (Pandas DataFrame)

   surface  rooms_new  zipcode_new  price_new   latitude  longitude
0    138.0        4.0         1060     420000  40.804672 -73.963420
1    130.0        5.0         1087     550000  52.355590   5.000561
2    116.0        5.0         1061     425000  52.373044   4.837568
3     92.0        5.0         1035     349511  52.416895   4.906767
4    127.0        4.0         1013    1050000  52.396789   4.876607

Scores (10 fold 'shuffled' cross-validation - Rsquared)

Random Forest Regression (n_estim=20, max_depth= None, max_feat=4} 0.866
Polynomial Regression (degrees=4) 0.810
Decision Tree Regression (max_depth=4, min_samples_leaf=6) 0.737
Neural Network MLP Regression (layer =[3,3], alpha=5, solv=lbfgs) 0.721
KNN Regression (n-neighbors = 15) 0.704
Ordinary Least-Squares Regression: 0.695
Ridge Regression (alpha = 0.1) 0.695
Support Vector Regression (kernel='linear', gamma = 0.001, C= 10) 0.690
Lasso Regression (alpha = 0.25) 0.614

Scatter plot - Surface vs. Asking Price (EUR)

OLS - Predicted prices vs. True price (EUR)

KNN Regression - Validation curve

Lasso Regression (Alpha = 0.25) - Predicted prices vs. True price (EUR)

Ridge Regression - Mean Test Score vs. Alpha

Polynomial Regression (degrees = 4) - Predicted prices vs. True price (EUR)

Random Forest Regression (n_estim=20, max_depth= None, max_feat=4) - Predicted prices vs. True price (EUR)

Support Vector Regression (kernel='linear', gamma = 0.001, C= 10) - Predicted prices vs. True price (EUR)

Neural Network MLP Regression (layer =[3,3], alpha=5, solv=lbfgs) (max_depth=4, min_samples_leaf=6) - Predicted prices vs. True price (EUR)

fannie-wong / predict_real_estate_prices Goto Github PK

predict_real_estate_prices's Introduction

Predicting Amsterdam house / real estate prices using Linear Regression, KNN-, Lasso-, Ridge-, Polynomial-, Support Vector (SVR)-, Decision Tree-, Random Forest-, and Neural Network MLP Regression.

Approach:

Packages required

Results along (Dec-17) Amsterdam house / real estate price data retrieved by means of the following scraper

Sample data input (Pandas DataFrame)

Scores (10 fold 'shuffled' cross-validation - Rsquared)

Scatter plot - Surface vs. Asking Price (EUR)

OLS - Predicted prices vs. True price (EUR)

KNN Regression - Validation curve

Lasso Regression (Alpha = 0.25) - Predicted prices vs. True price (EUR)

Ridge Regression - Mean Test Score vs. Alpha

Polynomial Regression (degrees = 4) - Predicted prices vs. True price (EUR)

Random Forest Regression (n_estim=20, max_depth= None, max_feat=4) - Predicted prices vs. True price (EUR)

Support Vector Regression (kernel='linear', gamma = 0.001, C= 10) - Predicted prices vs. True price (EUR)

Neural Network MLP Regression (layer =[3,3], alpha=5, solv=lbfgs) (max_depth=4, min_samples_leaf=6) - Predicted prices vs. True price (EUR)

predict_real_estate_prices's People

Contributors

Recommend Projects

Recommend Topics

Recommend Org