def create_features(self):
train_oht, test_oht = data.category_columns_to_one_hot_train_test(
X_train_origin, test_origin)
data.check_columns_size(train_oht, test_oht)
self.train = train_oht
self.test = test_oht
def create_features(self):
df_all_zero_per = data.get_zero_percent(df_all_origin)
train_zero_per = df_all_zero_per[:train_test_split_index]
test_zero_per = df_all_zero_per[train_test_split_index:]
test_zero_per = test_zero_per.reset_index(drop=True)
data.check_columns_size(train_zero_per, test_zero_per)
self.train = train_zero_per
self.test = test_zero_per
def create_features(self):
columns_list = ["KitchenQual", "GarageQual", "ExterQual", "BsmtQual"]
dict_list = [{
"Ex": 3,
"Gd": 2,
"TA": 1,
"Fa": -1,
"Po": -2,
np.nan: -3
}, {
"Ex": 3,
"Gd": 2,
"TA": 1,
"Fa": -1,
"Po": -2,
np.nan: -3
}, {
"Ex": 3,
"Gd": 2,
"TA": 1,
"Fa": -1,
"Po": -2,
np.nan: -3
}, {
"Ex": 3,
"Gd": 2,
"TA": 1,
"Fa": -1,
"Po": -2,
np.nan: -3
}]
df_new_train = pd.DataFrame(index=X_train_origin.index)
df_new_test = pd.DataFrame(index=test_origin.index)
for i in range(len(columns_list)):
df_new_train = pd.concat([
df_new_train,
data.ordinal_ordered_encoding(X_train_origin, columns_list[i],
dict_list[i])
],
axis=1)
for i in range(len(columns_list)):
df_new_test = pd.concat([
df_new_test,
data.ordinal_ordered_encoding(test_origin, columns_list[i],
dict_list[i])
],
axis=1)
data.check_columns_size(df_new_train, df_new_test)
self.train = df_new_train
self.test = df_new_test
def create_features(self):
column = "OverallQual"
df_target_train = pd.DataFrame(
index=train_origin.index,
columns=["{}".format(column) + "_target"])
for train_index, valid_index in kf.split(train_origin):
X_tr = train_origin.iloc[train_index, :]
X_va = train_origin.iloc[valid_index, :]
y_tr = Y_train_origin[train_index]
y_va = Y_train_origin[valid_index]
df_te = data.target_encoding(X_tr, X_va, column, y_va)
df_target_train.iloc[train_index, :] = df_te.values
df_target_test = data.target_encoding(test_origin, train_origin,
column, Y_train_origin)
data.check_columns_size(df_target_train, df_target_test)
self.train = df_target_train
self.test = df_target_test
def create_features(self):
all_bin = data.binning(df_all_origin, "YearBuilt", 10)
train_bin = all_bin[:train_test_split_index]
test_bin = all_bin[train_test_split_index:].reset_index(drop=True)
column = train_bin.columns[0]
df_target_train = pd.DataFrame(
index=train_bin.index, columns=["{}".format(column) + "_target"])
for train_index, valid_index in kf.split(train_bin):
X_tr = train_bin.iloc[train_index, :]
X_va = train_bin.iloc[valid_index, :]
y_tr = Y_train_origin[train_index]
y_va = Y_train_origin[valid_index]
df_te = data.target_encoding(X_tr, X_va, column, y_va)
df_target_train.iloc[train_index, :] = df_te.values
df_target_test = data.target_encoding(test_bin, train_bin, column,
Y_train_origin)
data.check_columns_size(df_target_train, df_target_test)
self.train = df_target_train
self.test = df_target_test
def create_features(self):
train_nan, test_nan = data.get_nan_flag(X_train_origin, test_origin)
data.check_columns_size(train_nan, test_nan)
self.train = train_nan
self.test = test_nan