scikit-learn Pipeline in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of a simple scikit-learn Pipeline

What is the output of the following code snippet when predicting with the pipeline?

ML Python

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
import numpy as np

pipeline = Pipeline([
    ('scaler', StandardScaler()),
    ('clf', LogisticRegression(random_state=0))
])

X_train = np.array([[1, 2], [2, 3], [3, 4], [4, 5]])
y_train = np.array([0, 0, 1, 1])

pipeline.fit(X_train, y_train)

X_test = np.array([[1.5, 2.5]])
prediction = pipeline.predict(X_test)
print(prediction)

A [0 1]

B [0]

C [1]

D Raises a ValueError

💡 Hint

Think about how the pipeline transforms data before prediction.

The pipeline first scales the input using StandardScaler, then applies LogisticRegression. Given the training data, the model predicts class 1 for the test input.

❓ Model Choice

intermediate

Choosing the correct pipeline step for text data

You want to build a pipeline to classify text documents. Which step should you include before the classifier to convert text into numbers?

A KMeans()

B CountVectorizer()

C PCA()

D StandardScaler()

💡 Hint

Text data needs to be converted into numeric features before classification.

CountVectorizer converts text documents into a matrix of token counts, which can be used by classifiers.

❓ Hyperparameter

advanced

Setting hyperparameters in a pipeline

Given this pipeline: pipeline = Pipeline([ ('scaler', StandardScaler()), ('clf', LogisticRegression()) ]) How do you set the LogisticRegression parameter 'C' to 0.5 when using GridSearchCV?

A {'scaler__C': [0.5]}

B {'C': [0.5]}

C {'pipeline__C': [0.5]}

D {'clf__C': [0.5]}

💡 Hint

Use double underscores to access parameters of steps inside the pipeline.

Parameters of steps in a pipeline are accessed by 'stepname__parameter'. Here, 'clf' is the step name.

❓ Metrics

advanced

Evaluating pipeline performance with cross-validation

You run cross_val_score on a pipeline with a classifier and get these scores: [0.8, 0.85, 0.78, 0.82, 0.81]. What is the mean accuracy?

A 0.81

B 0.82

C 0.80

D 0.83

💡 Hint

Add all scores and divide by the number of scores.

Mean accuracy = (0.8 + 0.85 + 0.78 + 0.82 + 0.81) / 5 = 0.812, rounded to 0.81.

🔧 Debug

expert

Identifying error in pipeline usage

What error does this code raise? from sklearn.pipeline import Pipeline from sklearn.preprocessing import StandardScaler from sklearn.linear_model import LogisticRegression pipeline = Pipeline([ ('scaler', StandardScaler()), ('clf', LogisticRegression()) ]) X_test = [[1, 2], [3, 4]] prediction = pipeline.predict(X_test)

A ValueError: This Pipeline instance is not fitted yet. Call 'fit' with appropriate arguments before using this estimator.

B AttributeError: 'Pipeline' object has no attribute 'predict'

C TypeError: 'list' object is not callable

D No error, outputs predictions

💡 Hint

You must train the pipeline before predicting.

The pipeline is not fitted before calling predict, so it raises a ValueError.