def test_binary(self):
print(bin_train_df.head())
drop_list = []
numeric_list = []
cat_list = []
xgb = xgbmagic.Xgb(bin_train_df, target_column='y', id_column='id', numeric_columns=numeric_list, drop_columns=drop_list, categorical_columns=cat_list, num_training_rounds=100, target_type='binary', verbose=True, prefix='test', sample_fraction=0.3, n_samples=2)
xgb.train()
output = xgb.predict(bin_test_df)
print('OUTPUT', output)
self.assertTrue(len(output) == len(bin_test_dict['id']))
def test_random_sample(self):
n_samples = 3
fraction = 0.2
drop_list = []
numeric_list = []
cat_list = []
xgb = xgbmagic.Xgb(train_df, target_column='y', id_column='id', numeric_columns=numeric_list, drop_columns=drop_list, categorical_columns=cat_list, num_training_rounds=100, target_type='linear', verbose=True, prefix='test', sample_fraction=0.01, n_samples=2)
samples = xgb.random_sample(train_df, fraction=fraction, n_samples=n_samples)
self.assertTrue(len(samples)==n_samples)
self.assertTrue(len(samples[0])==round(fraction*len(train_df)))
def test_output(self):
drop_list = []
numeric_list = []
cat_list = []
xgb = xgbmagic.Xgb(train_df, target_column='y', id_column='id', numeric_columns=numeric_list, drop_columns=drop_list, categorical_columns=cat_list, num_training_rounds=100, target_type='linear', verbose=True, prefix='test', sample_fraction=0.3, n_samples=2)
xgb.train()
output = xgb.predict(test_df)
print('OUTPUT', output)
self.assertTrue(len(output) == len(test_dict['id']))
multi_outputs = xgb.predict(test_df, return_multi_outputs=True)
self.assertTrue(len(multi_outputs) == xgb.n_samples)
def test_init(self):
drop_list = []
numeric_list = []
cat_list = []
xgb = xgbmagic.Xgb(train_df,
target_column='y',
id_column='id',
numeric_columns=numeric_list,
drop_columns=drop_list,
categorical_columns=cat_list,
num_training_rounds=100,
target_type='linear',
verbose=True,
prefix='test',
sample_fraction=0.01,
n_samples=2)
self.assertIsInstance(xgb, xgbmagic.Xgb)
import xgbmagic
import pandas as pd
import pickle
# train and test sets taken from Kaggle's Santander Customer Satisfaction challenge
df = pd.read_csv('train.csv')
xgb = xgbmagic.Xgb(df,
target_column='TARGET',
id_column='ID',
categorical_columns=[],
num_training_rounds=1000,
target_type='binary',
early_stopping_rounds=50)
xgb.train()
test_df = pd.read_csv('test.csv')
print(xgb.feature_importance())
output = xgb.predict(test_df)
xgb.write_csv('output-xgbmagic.csv')
import xgbmagic
import pandas as pd
from sklearn.model_selection import StratifiedKFold, KFold, cross_val_score
df = pd.read_csv(
'/home/rlougee/Desktop/invitrodb_v2_enrichments/CTEW_aeid_100_invitrodbv2_20180918/CTEW_Results/CT-Enriched_FP_aeid_100_invitrodbv2_20180918.tsv',
delimiter='\t')
df.iloc[:, 2:] = df.iloc[:, 2:].astype(int)
for i in df.columns:
f = i
f = f.replace('[', '')
f = f.replace(']', '')
f = f.replace('<', '')
if i != f:
df = df.rename(columns={i: f})
# classification model = binary
target_type = 'binary'
# set columns type here
xgb = xgbmagic.Xgb(df.iloc[50:, :],
target_column='Hit_Call',
id_column='Dsstox_Compound_ID',
numeric_columns=df.columns[2:])
xgb.train()
print(xgb.feature_importance())
print(xgb.predict(df.iloc[:50, :]))