MBTest

Mercedes-Benz Test Bench Optimization-Based Question

Overview

This project focuses on reducing the time a Mercedes-Benz car spends on the test bench. Using historical data from Mercedes-Benz, we build a model to predict the duration of the test phase. This repository contains the data preprocessing and machine learning steps taken to develop an XGBoost regression model capable of making these predictions.

Dataset

The dataset comprises two files:

• train.csv: Contains the training data with features and the target variable.
• test.csv: Contains the test data without the target variable.

The features include categorical and binary data that describe different permutations of Mercedes-Benz car features.

Prerequisites

Before running the project, ensure that you have the following requirements installed:

• Python 3.8+
• Pandas library
• scikit-learn library
• XGBoost library

You can install the dependencies with:

pip install pandas scikit-learn xgboost

Usage

To run the model and generate predictions, follow these steps:

• Place your train.csv and test.csv in the same directory as the mercedes_benz_optimization.py script.
• Execute the script:

  python mercedes_benz_optimization.py

• After running, the script will generate a file named mercedes_benz_test_predictions.csv with the predictions.

Model Development

The script mercedes_benz_optimization.py includes the following steps:

• Data preprocessing:
  1. Removal of features with zero variance
  2. One-hot encoding of categorical variables
  3. Dimensionality reduction with PCA (retaining 95% variance)
• Hyperparameter tuning of the XGBoost regressor with grid search and cross-validation
• Training the best model and making predictions on the test set

Output

The output will be a CSV file with the predicted test times for each ID in the test dataset.

Authors

Dennis Sagayanathan