def centerize_dataframe(input_dataframe, label_name: str or list):
"""
# Apply centering to data frame
:param input_dataframe: input pandas data frame
:param label_name: exception column
:return centerized_dataframe: normalized data frame
:return: mean_list: list of mean
"""
# Make a copy of the data frame
centerized_dataframe = input_dataframe.copy()
# Extract only data part
data = centerized_dataframe[centerized_dataframe.columns[
centerized_dataframe.columns != label_name]]
# Replace NaN/Inf as 0
clean_data = DataCleaner.clean_data(data)
# Apply centering to all columns
mean_list = []
for feature_name in data.columns:
mean_value = clean_data[feature_name].mean()
centerized_dataframe[
feature_name] = clean_data[feature_name] - mean_value
mean_list.append(mean_value)
# Put back label column
centerized_dataframe[label_name] = input_dataframe[label_name]
return centerized_dataframe, mean_list
def standardize_dataframe(input_dataframe, label_name: str or list):
"""
# Apply standardization
:param input_dataframe: input pandas data frame
:param label_name: exception column
:return standardized_dataframe: standardized data frame
:return std_list: list of standard deviation
"""
# Make a copy of the data frame
standardized_dataframe = input_dataframe.copy()
# Extract only data part
data = standardized_dataframe[standardized_dataframe.columns[
standardized_dataframe.columns != label_name]]
# Clean data
clean_data = DataCleaner.clean_data(data)
# Apply standardization to all columns
std_list = []
for feature_name in clean_data.columns:
std_value = np.std(clean_data[feature_name])
standardized_dataframe[feature_name] = (
clean_data[feature_name] -
np.mean(clean_data[feature_name])) / std_value
std_list.append(std_value)
# Put back label column
standardized_dataframe[label_name] = input_dataframe[label_name]
return standardized_dataframe, std_list
def normalize_dataframe(input_dataframe, label_name: str or list):
"""
# Apply normalization to data frame
:param input_dataframe: input pandas data frame
:param label_name: exception column
:return normalized_dataframe: normalized data frame
"""
# Make a copy of the data frame
normalized_dataframe = input_dataframe.copy()
# Extract only data part
data = normalized_dataframe[normalized_dataframe.columns[
normalized_dataframe.columns != label_name]]
# Replace NaN/Inf as 0
clean_data = DataCleaner.clean_data(data)
# Apply normalizaion to all columns
for feature_name in data.columns:
max_value = clean_data[feature_name].max()
min_value = clean_data[feature_name].min()
normalized_dataframe[feature_name] = (
clean_data[feature_name] - min_value) / (max_value - min_value)
# Put back label column
normalized_dataframe[label_name] = input_dataframe[label_name]
return normalized_dataframe