Create a Complete ML Pipeline

Now let's build a proper pipeline with the final estimator. As a result, we will get a trained prediction pipeline that can be used for predicting new instances simply by calling the .predict() method.
To train a predictor (model), you need y to be encoded. This is done separately from the Pipeline we build for X.
Remember that LabelEncoder is used for encoding the target.

You have the same Penguins dataset.
The task is to build a pipeline with KNeighborsClassifier as a final estimator, train it, and predict values for the X itself.
Since the predictions will be encoded (0, 1, or 2), we will use the .inverse_transform() method of LabelEncoder to get the predictions back to 'Adelie', 'Chinstrap', or 'Gentoo'.

Aufgabe

Swipe to start coding

Encode the y variable using the LabelEncoder.
Make a pipeline containing ct, SimpleImputer, StandardScaler, and KNeighborsClassifier (in that order).
Train the pipe object.

Lösung

import pandas as pd

from sklearn.impute import SimpleImputer

from sklearn.preprocessing import OneHotEncoder, LabelEncoder, StandardScaler

from sklearn.compose import make_column_transformer

from sklearn.pipeline import make_pipeline

from sklearn.neighbors import KNeighborsClassifier

df = pd.read_csv('https://codefinity-content-media.s3.eu-west-1.amazonaws.com/a65bbc96-309e-4df9-a790-a1eb8c815a1c/penguins.csv')

# Removing rows with more than 1 null

df = df[df.isna().sum(axis=1) < 2]

# Assigining X, y variables

X, y = df.drop('species', axis=1), df['species']

# Encode the target

label_enc = LabelEncoder()

y = label_enc.fit_transform(y)

# Create the ColumnTransformer for encoding features

ct = make_column_transformer((OneHotEncoder(), ['island', 'sex']),

remainder='passthrough')

# Make a Pipeline of ct, SimpleImputer, and StandardScaler

pipe = make_pipeline(ct,

SimpleImputer(strategy='most_frequent'),

StandardScaler(),

KNeighborsClassifier()

)

# Train the model

pipe.fit(X, y)

# Print predictions

y_pred = pipe.predict(X) # Get encoded predictions

print(label_enc.inverse_transform(y_pred)) # Decode predictions and print

War alles klar?

Danke für Ihr Feedback!

Abschnitt 3. Kapitel 6