normalize_transform_operator¶
Operator to normalize the integer/float values in a dataframe
-
class
tasrif.processing_pipeline.custom.normalize_transform_operator.NormalizeTransformOperator(feature_names='all', model=None)¶ Normalizes the values for the supplied feature_names using the supplied model. This operator works on a 2D data frames where the feature_names represent the features. The returned data frame contains normalized values in the specified feature_names. This will be used when splitting the dataset into training and testing set and then using the model generated from the testing set on the testing set.
Examples
>>> import pandas as pd >>> from sklearn.model_selection import train_test_split >>> from tasrif.processing_pipeline.custom import NormalizeOperator >>> from tasrif.processing_pipeline.custom import NormalizeTransformOperator
>>> df = pd.DataFrame([ [1, "2020-05-01 00:00:00", 10], [1, "2020-05-01 01:00:00", 15], [1, "2020-05-01 03:00:00", 23], [2, "2020-05-02 00:00:00", 17], [2, "2020-05-02 01:00:00", 11]], columns=['logId', 'timestamp', 'sleep_level'])
>>> X_train, X_test, y_train, y_test = train_test_split(df['timestamp'], df['sleep_level'], test_size=0.4)
>>> op1 = NormalizeOperator('all', 'minmax', {'feature_range': (0, 2)})
>>> output1 = op1.process(y_train.to_frame())
>>> print(output1)
- [(array([[2. ],
[1.33333333], [0. ]]), MinMaxScaler(feature_range=(0, 2)))]
>>> processed_train_y = output1[0][0] >>> trained_model = output1[0][1]
>>> op2 = NormalizeTranformOperator('all', trained_model)
>>> output2 = op2.process(y_test.to_frame())
>>> print(output2)
- [array([[ 4. ],
[-0.33333333]])]
-
__init__(feature_names='all', model=None)¶ Creates a new instance of NormalizeTransformOperator
- Parameters
feature_names (list, str) – feature_names in the given dataframe to normalize
model (StandardScaler / MinMaxScaler / MaxAbsScaler / RobustScaler) – Model with normalization method (‘zscore’, ‘minmax’, ‘maxabs’, ‘robust’)
- Raises
ValueError – parameter method unknown.