Recursive Feature Elimination to Select Key Process Variables for Assembly Line Productivity Prediction

Project info

• Field of studyIndustrial Engineering and Production
• LevelMaster's thesis (S2)
• Method typeDimensionality Reduction
• Estimated build time± 5 working days
• Toolsscikit-learn RFECV, RFE, LogisticRegression
• PriceUS$22 - US$147

Research goal

Reduce the dimensions of number of operators and setup time related to line productivity using Recursive Feature Elimination to summarise the main pattern.

Variables

Example result

• 5Optimal features
• 5Features removed
• 0.932CV R2 (best)
• 0.951R2 (test)
• 3.582RMSE (test)
• 1True recall

RFE keeps 5 of 10 predictors for Line productivity with a cross-validated R2 of 0.932 and a test R2 of 0.951.

Computed from illustrative synthetic data, not from a real study.

About Recursive Feature Elimination

Recursive Feature Elimination removes the least important features step by step based on model weights until the desired number remains. It suits selecting feature subsets for any model that exposes coefficients or importances.

Steps

1. 1Standardise features and split into training and test sets
2. 2Choose a base estimator (for example logistic regression)
3. 3Run RFECV with cross-validation
4. 4Inspect the optimal feature count and feature ranking
5. 5Refit with the selected features and evaluate on the test set
6. 6Plot the validation score against number of features

Assumptions to check

• The base estimator provides weights or importances
• Selection is done on training data only
• Correlated features can substitute each other so rankings are not unique

Reading and reporting the results

Report the optimal number of features, the selected list, and the test score against the full model. State the base estimator used because it influences the ranking.

Example Python code

A short starting script with this project's variable names. The delivered version is adapted to your data and fully commented.

import numpy as np
import matplotlib.pyplot as plt
from sklearn.datasets import make_classification
from sklearn.feature_selection import RFECV
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split, StratifiedKFold
from sklearn.metrics import accuracy_score

X, line_productivity = make_classification(n_samples=400, n_features=15, n_informative=4,
                               n_redundant=2, random_state=24)
number_of_operators = X[:, 0]
setup_time = X[:, 1]

X_tr, X_te, y_tr, y_te = train_test_split(X, line_productivity, test_size=0.25, stratify=line_productivity, random_state=24)
sc = StandardScaler().fit(X_tr)
Xs_tr, Xs_te = sc.transform(X_tr), sc.transform(X_te)

est = LogisticRegression(max_iter=1000)
rfecv = RFECV(est, step=1, cv=StratifiedKFold(5), scoring="accuracy").fit(Xs_tr, y_tr)
print("Optimal number of features:", rfecv.n_features_)
print("Selected:", np.flatnonzero(rfecv.support_))

acc_sel = accuracy_score(y_te, rfecv.predict(Xs_te))
acc_full = accuracy_score(y_te, est.fit(Xs_tr, y_tr).predict(Xs_te))
print("Test accuracy selected:", round(acc_sel, 3), "full:", round(acc_full, 3))

scores = rfecv.cv_results_["mean_test_score"]
plt.plot(range(1, len(scores) + 1), scores, marker="o")
plt.xlabel("Number of features"); plt.ylabel("CV accuracy")
plt.show()

Packages and prices

Choose how much help you need. Each higher package includes everything in the lower ones. Prices exclude 11% VAT.

• Payment: 50% down payment before work starts, balance after the deliverables are handed over
• Tax: Prices exclude 11% VAT
• Data and academic integrity: Results are computed from your own research data; you write and stand behind your manuscript, we help with the analysis and your understanding of it
• Revisions: Up to 2 revisions are included in the price (Package 3: 3 rounds)

Frequently asked questions