XGBoost

import xgboost as xgb
from sklearn import metrics

def auc(m, train, test): 
    return (metrics.roc_auc_score(y_train,m.predict_proba(train)[:,1]),
                            metrics.roc_auc_score(y_test,m.predict_proba(test)[:,1]))

# Parameter Tuning
model = xgb.XGBClassifier()
param_dist = {"max_depth": [10,30,50],
              "min_child_weight" : [1,3,6],
              "n_estimators": [200],
              "learning_rate": [0.05, 0.1,0.16],}
grid_search = GridSearchCV(model, param_grid=param_dist, cv = 3, 
                                   verbose=10, n_jobs=-1)
grid_search.fit(train, y_train)

grid_search.best_estimator_

model = xgb.XGBClassifier(max_depth=50, min_child_weight=1,  n_estimators=200,\
                          n_jobs=-1 , verbose=1,learning_rate=0.16)
model.fit(train,y_train)

auc(model, train, test)

Hi Alvira,

Read your awesome post about xgboost/lightgbm/catboost on Medium... coming here hoping to ask you a couple of questions

1. I simply copied&pasted&ran your code (lightgbm part), and turned out if I ran model2.predict(train), the predictions are real numbers instead of binary numbers. Aren’t we suppose to give binary predictions (delayed or not)?
2. If real number predictions are intended, is it because you need real numbers to plot AUC? (I’m going to read more about AUC tonight) And why don’t we use binary preditions to plot AUC?

Thank you in advance!

Best,
JJ

Great example of a grid search with custom metric.

Generally, you can use the grid_search.best_estimator_ property to access a fit model directly. No need to re-train a model.