Sklearn - Pipelines




1. Pipeline 1




Création du pipeline 1
from sklearn.preprocessing import PolynomialFeatures, MinMaxScaler, OneHotEncoder, StandardScaler
from sklearn.feature_selection import SelectKBest, f_regression
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline, make_pipeline
pipeline1 = Pipeline(steps=[
  ('Features engineering', PolynomialFeatures()),
  ('Feature Selection', SelectKBest()),
  ('Normaliser', MinMaxScaler())
])
Pipeline(steps=[('Features engineering', PolynomialFeatures()),
                ('Feature Selection', SelectKBest()),
                ('Normaliser', MinMaxScaler())])



2. Pipeline 2




Création du Column Transformer
categorial_col = []
column_transformer = ColumnTransformer(transformers=[
  ('Encoder', OneHotEncoder(), categorial_col)
],
remainder='passthrough'
)
ColumnTransformer(remainder='passthrough',
                  transformers=[('Encoder', OneHotEncoder(), [])])

Création du pipeline 2
pipeline2 = Pipeline(steps=[
  ('Encoder', column_transformer),
  ('Feature engineering', PolynomialFeatures()),
  ('Normaliser', MinMaxScaler())
])
Pipeline(steps=[('Encoder',
                 ColumnTransformer(remainder='passthrough',
                                   transformers=[('Encoder', OneHotEncoder(),
                                                  [])])),
                ('Feature engineering', PolynomialFeatures()),
                ('Normaliser', MinMaxScaler())])



3. Pipeline 3




Création du Pipeline des catégories
category_pipeline = Pipeline(steps=[
    ('Encoder', OneHotEncoder()),
    ('Select', SelectKBest(score_func=f_regression, k=3))
])
Pipeline(steps=[('Encoder', OneHotEncoder()),
                ('Select',
                 SelectKBest(k=3,
                             score_func=))])

Création du Pipeline des catégories
categorial_col = []
column_transformer2 = ColumnTransformer(transformers=[
        ('Cat Pipeline', category_pipeline, categorial_col)
    ],
    remainder='passthrough'
)
ColumnTransformer(remainder='passthrough',
                  transformers=[('Cat Pipeline',
                                 Pipeline(steps=[('Encoder', OneHotEncoder()),
                                                 ('Select',
                                                  SelectKBest(k=3,
                                                              score_func=))]),
                                 [])])

Création du Pipeline 3
pipeline3 = make_pipeline(
  column_transformer2,
  PolynomialFeatures(),
  MinMaxScaler()
)
Pipeline(steps=[('columntransformer',
                 ColumnTransformer(remainder='passthrough',
                                   transformers=[('Cat Pipeline',
                                                  Pipeline(steps=[('Encoder',
                                                                   OneHotEncoder()),
                                                                  ('Select',
                                                                   SelectKBest(k=3,
                                                                               score_func=))]),
                                                  [])])),
                ('polynomialfeatures', PolynomialFeatures()),
                ('minmaxscaler', MinMaxScaler())])



4. Pipeline 4




Création du Pipeline categoriel
categorial_col = []
continuous_col = ['height', 'weight']discret_col = ['age', 'floor']cat_tranformer2 = Pipeline(steps=[('encoder', OneHotEncoder()), ('selecter', SelectKBest(score_func=f_regression, k=5))])
Pipeline(steps=[('encoder', OneHotEncoder()),
                ('selecter',
                 SelectKBest(k=5,
                             score_func=))])

Création du Column transformer
column_transformer3 = ColumnTransformer(transformers=[
   ('Encode Select', cat_tranformer2, categorial_col),
   ('Standardizer', StandardScaler(), continuous_col),
   ('MinMax', MinMaxScaler(), discret_col)
])
ColumnTransformer(transformers=[('Encode Select',
                                 Pipeline(steps=[('encoder', OneHotEncoder()),
                                                 ('selecter',
                                                  SelectKBest(k=5,
                                                              score_func=))]),
                                 []),
                                ('Standardizer', StandardScaler(),
                                 ['height', 'weight']),
                                ('MinMax', MinMaxScaler(), ['age', 'floor'])])

Création du Pipeline4
pipeline4 = Pipeline(steps=[
    ('Encode_Select_Scale', column_transformer3),
    ('Feature Engineering', PolynomialFeatures()),
    ('MinMax', MinMaxScaler())
])
Pipeline(steps=[('Encode_Select_Scale',
                 ColumnTransformer(transformers=[('Encode Select',
                                                  Pipeline(steps=[('encoder',
                                                                   OneHotEncoder()),
                                                                  ('selecter',
                                                                   SelectKBest(k=5,
                                                                               score_func=))]),
                                                  []),
                                                 ('Standardizer',
                                                  StandardScaler(),
                                                  ['height', 'weight']),
                                                 ('MinMax', MinMaxScaler(),
                                                  ['age', 'floor'])])),
                ('Feature Engineering', PolynomialFeatures()),
                ('MinMax', MinMaxScaler())])