catboost

catboost.py

import os
import pandas as pd
import numpy as np
np.set_printoptions(precision=4)
import catboost
print(catboost.__version__)

#dataset

from catboost.datasets import amazon
(train_df, test_df) = amazon()

y = train_df.ACTION
X = train_df.drop('ACTION', axis=1)

#all the features are categorical

cat_features = list(range(0, X.shape[1]))
print(cat_features)

#unbalanced labels

print('Labels: {}'.format(set(y)))
print('Zero count = {}, One count = {}'.format(len(y) - sum(y), sum(y)))

#training
from catboost import CatBoostClassifier
model = CatBoostClassifier(iterations=100)
model.fit(X, y, cat_features=cat_features, verbose=10)

model.predict_proba(X)

from catboost import Pool
pool = Pool(data=X, label=y, cat_features=cat_features)

from sklearn.model_selection import train_test_split

data = train_test_split(X, y, test_size=0.2, random_state=0)
X_train, X_validation, y_train, y_validation = data

train_pool = Pool(
    data=X_train, 
    label=y_train, 
    cat_features=cat_features
)

validation_pool = Pool(
    data=X_validation, 
    label=y_validation, 
    cat_features=cat_features
)

model = CatBoostClassifier(
    iterations=5,
    learning_rate=0.1,
    # loss_function='CrossEntropy'
)
model.fit(train_pool, eval_set=validation_pool, verbose=False)

print('Model is fitted: {}'.format(model.is_fitted()))
print('Model params:\n{}'.format(model.get_params()))

model = CatBoostClassifier(
    iterations=15,
#     verbose=5,
)
model.fit(train_pool, eval_set=validation_pool);

model = CatBoostClassifier(
    iterations=50,
    learning_rate=0.5,
    custom_loss=['AUC', 'Accuracy']
)

model.fit(
    train_pool,
    eval_set=validation_pool,
    verbose=False,
    plot=True
);

model_with_early_stop = CatBoostClassifier(
    iterations=200,
    learning_rate=0.5,
    early_stopping_rounds=20
)

model_with_early_stop.fit(
    train_pool,
    eval_set=validation_pool,
    verbose=False,
    plot=True
);


from catboost import cv

params = {
    'loss_function': 'Logloss',
    'iterations': 80,
    'custom_loss': 'AUC',
    'learning_rate': 0.5,
}

cv_data = cv(
    params = params,
    pool = train_pool,
    fold_count=5,
    shuffle=True,
    partition_random_seed=0,
    plot=True,
    verbose=False
)

cv_data.head(10)

from sklearn.model_selection import GridSearchCV

param_grid = {
    "iterations": [10,100],
    "learning_rate": [0.01,0.1],
    "depth": [4,7],
    "early_stopping_rounds" : [5,10],
    "depth" : [4,8],
    "l2_leaf_reg": [2,4]
}

clf = CatBoostClassifier(
    cat_features=cat_features, 
    verbose=20
)
grid_search = GridSearchCV(clf, param_grid=param_grid, cv=3)
results = grid_search.fit(X_train, y_train)
results.best_estimator_.get_params()