Recursive feature elimination in ML Python - Model Metrics & Evaluation

Metrics & Evaluation - Recursive feature elimination

Which metric matters for Recursive Feature Elimination and WHY

Recursive Feature Elimination (RFE) helps pick the best features for a model. The key metric to watch is the model's performance metric like accuracy, F1 score, or mean squared error after each feature removal step. This shows if removing features helps or hurts the model. We want to keep features that improve or keep performance stable.

Confusion Matrix Example

Imagine a classification model using RFE. After selecting features, we check the confusion matrix:

      | Predicted Positive | Predicted Negative |
      |--------------------|--------------------|
      | True Positive (TP): 50 | False Negative (FN): 10 |
      | False Positive (FP): 5 | True Negative (TN): 35 |
    

Total samples = 50 + 10 + 5 + 35 = 100

Precision = 50 / (50 + 5) = 0.91

Recall = 50 / (50 + 10) = 0.83

F1 Score = 2 * (0.91 * 0.83) / (0.91 + 0.83) ≈ 0.87

These metrics tell us how well the model performs with the chosen features.

Precision vs Recall Tradeoff in Feature Selection

When RFE removes features, it can affect precision and recall differently.

• High Precision Needed: For spam detection, we want to avoid marking good emails as spam. So, RFE should keep features that help precision.
• High Recall Needed: For disease detection, missing a sick patient is bad. RFE should keep features that help recall.

RFE helps find the smallest feature set that balances these metrics well.

Good vs Bad Metric Values for RFE

Good: After RFE, model accuracy or F1 score stays high or improves. For example, accuracy above 90% with fewer features means success.

Bad: Metrics drop a lot after removing features, like accuracy falling from 90% to 70%. This means important features were removed.

Common Pitfalls in Metrics with RFE

• Overfitting: If RFE is done on the whole dataset before splitting, it leaks information and inflates metrics.
• Ignoring Validation: Only checking training accuracy can mislead. Always check metrics on unseen data.
• Accuracy Paradox: High accuracy can hide poor recall or precision if classes are imbalanced.

Self Check

Your model after RFE has 98% accuracy but only 12% recall on fraud cases. Is it good?

Answer: No. Even with high accuracy, the model misses most fraud cases (low recall). For fraud detection, recall is critical to catch fraud. So, this model is not good for production.

Key Result

RFE success is measured by stable or improved model performance metrics (accuracy, F1) after feature removal, ensuring important features remain.

Practice

1. What is the main goal of Recursive Feature Elimination (RFE) in machine learning?

easy

A. To select the most important features by removing less important ones step by step

B. To increase the number of features in the dataset

C. To randomly shuffle the features before training

D. To create new features by combining existing ones

Solution

1. Step 1: Understand the purpose of RFE

   RFE works by removing less important features one at a time to keep only the best ones.

2. Step 2: Compare options to the purpose

   Only To select the most important features by removing less important ones step by step describes this step-by-step removal of less important features.

3. Final Answer:

   To select the most important features by removing less important ones step by step -> Option A

4. Quick Check:

   RFE = Stepwise feature removal [OK]

Hint: RFE removes features stepwise to keep the best ones [OK]

Common Mistakes:

• Thinking RFE adds or creates features
• Confusing RFE with random feature shuffling
• Believing RFE increases feature count

2. Which of the following is the correct way to import Recursive Feature Elimination from scikit-learn in Python?

easy

A. from sklearn.feature_selection import RecursiveFeatureElimination

B. from sklearn.feature_selection import RFE

C. import sklearn.feature_selection.RFE as rfe

D. from sklearn.selection import RFE

Solution

1. Step 1: Recall the correct import statement

   The class is named RFE and is in sklearn.feature_selection.

2. Step 2: Match options with correct syntax

   from sklearn.feature_selection import RFE correctly imports RFE from sklearn.feature_selection.

3. Final Answer:

   from sklearn.feature_selection import RFE -> Option B

4. Quick Check:

   Correct import is 'from sklearn.feature_selection import RFE' [OK]

Hint: Remember: RFE is imported directly from sklearn.feature_selection [OK]

Common Mistakes:

• Using wrong module name like sklearn.selection
• Trying to import full name RecursiveFeatureElimination
• Using incorrect import syntax

3. Given the following Python code using RFE, what will be the output of print(selected_features)?

from sklearn.datasets import load_iris
from sklearn.linear_model import LogisticRegression
from sklearn.feature_selection import RFE

iris = load_iris()
X, y = iris.data, iris.target
model = LogisticRegression(max_iter=200)
rfe = RFE(model, n_features_to_select=2)
rfe.fit(X, y)
selected_features = rfe.support_
print(selected_features)

medium

A. [ True True False False ]

B. [False True False True ]

C. [ True False True False ]

D. [False False True True ]

Solution

1. Step 1: Understand RFE output support_

   The support_ attribute is a boolean array showing which features are selected.

2. Step 2: Run RFE with LogisticRegression on iris dataset

   RFE selects the two most important features, which for iris are the last two features (petal length and petal width), so the output is [False False True True].

3. Final Answer:

   [False False True True ] -> Option D

4. Quick Check:

   RFE selects last two iris features = [False False True True] [OK]

Hint: Iris important features are last two; RFE selects those [OK]

Common Mistakes:

• Assuming first two features are selected
• Confusing support_ with ranking_
• Not setting max_iter causing convergence warnings

4. Identify the error in this RFE usage code and choose the correct fix:

from sklearn.feature_selection import RFE
from sklearn.linear_model import LogisticRegression

model = LogisticRegression()
rfe = RFE(model, n_features_to_select=0)
rfe.fit(X, y)

medium

A. n_features_to_select cannot be zero; set it to a positive integer

B. LogisticRegression must be imported from sklearn.linear_model.linear_model

C. RFE requires a random_state parameter

D. fit method requires sample_weight argument

Solution

1. Step 1: Check parameter n_features_to_select

   This parameter must be at least 1 or None, zero is invalid.

2. Step 2: Identify correct fix

   Setting n_features_to_select to a positive integer fixes the error.

3. Final Answer:

   n_features_to_select cannot be zero; set it to a positive integer -> Option A

4. Quick Check:

   n_features_to_select > 0 required [OK]

Hint: n_features_to_select must be positive, never zero [OK]

Common Mistakes:

• Setting n_features_to_select to zero
• Wrong import paths for LogisticRegression
• Thinking random_state is mandatory for RFE

5. You have a dataset with 20 features and want to use RFE with a Random Forest model to select the top 5 features. Which of the following code snippets correctly applies RFE and outputs the names of the selected features assuming your data is in a pandas DataFrame df and target in y?

hard

A. from sklearn.feature_selection import RFE
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier()
rfe = RFE(model, n_features_to_select=5)
rfe.fit(df, y)
selected = df.columns[rfe.ranking_ <= 5]
print(selected.tolist())

B. from sklearn.feature_selection import RFE
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier()
rfe = RFE(model, n_features_to_select=5)
rfe.fit(y, df)
selected = df.columns[rfe.support_]
print(selected.tolist())

C. from sklearn.feature_selection import RFE
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier()
rfe = RFE(model, n_features_to_select=5)
rfe.fit(df, y)
selected = df.columns[rfe.support_]
print(selected.tolist())

D. from sklearn.feature_selection import RFE
from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier()
rfe = RFE(model, n_features_to_select=5)
rfe.fit(df, y)
selected = df.columns[rfe.ranking_ == 5]
print(selected.tolist())

Solution

1. Step 1: Check correct fit method usage

   Features (df) must be first argument, target (y) second in fit.

2. Step 2: Select features using support_ boolean mask

   Use rfe.support_ to get selected features, then map to column names.

3. Final Answer:

   Code snippet A correctly fits and selects features using support_ mask -> Option C

4. Quick Check:

   fit(df, y) + support_ mask = correct feature selection [OK]

Hint: fit(df, y) and use support_ to get selected features [OK]

Common Mistakes:

• Swapping X and y in fit method
• Using ranking_ == 5 instead of support_
• Not converting boolean mask to column names