🔧 Predictive Maintenance System Using XGBoost and Random Forest 🚀

📝 Overview

This project implements a Predictive Maintenance System for industrial motors using two machine learning models: XGBoost and Random Forest. The models predict the type of failure that may occur in the motors, such as:

• ⚡ Power Failure
• 🛠️ Tool Wear Failure
• 💥 Overstrain Failure

The predictions are made based on sensor data like:

• 🌡️ Air temperature
• 🌡️ Process temperature
• 🔄 Rotational speed
• ⚙️ Torque
• ⏳ Tool wear

🛠️ Installation

To set up the project, you need to install the required libraries. Run the following commands:

!pip install pandas numpy matplotlib seaborn plotly
!pip install xgboost scikit-learn

📊 Dataset Overview

The dataset contains sensor readings from motors with the following key features:

• 🌡️ Air temperature [K]
• 🌡️ Process temperature [K]
• 🔄 Rotational speed [rpm]
• ⚙️ Torque [Nm]
• ⏳ Tool wear [min]
• 🎯 Failure Type (Target variable with 6 classes)

🛠️ Project Workflow

1. ⚙️ Data Preprocessing:

• Load and clean the dataset.
• Drop unnecessary columns like Product ID, UDI, and any column with more than 5% missing values.
• Rename columns for clarity (e.g., Air temperature [K] → air_temperature).

2. 📊 Exploratory Data Analysis (EDA):

• 📈 Generate histograms and pairplots for visualizing relationships between sensor readings and failure types.
• 🔍 Compute and visualize correlations between numerical features.

3. 🤖 Modeling:

• Two models are trained:

  • 🌲 XGBoost: A tree-based algorithm optimized with the gpu_hist method.
  • 🌳 Random Forest: A classic ensemble method using 100 trees.

• Cross-validation is performed for the XGBoost model, and performance metrics such as accuracy, confusion matrices, and classification reports are generated.

4. 🛠️ Manual Predictions:

• A function allows users to input sensor values manually for real-time predictions of motor failures.

5. ⚖️ Model Comparison:

• The accuracies of XGBoost and Random Forest are compared, and the model with the highest accuracy is identified.

6. 💾 Saving & Loading Models:

• Models are saved as .pkl files for future use.

7. 📊 Predictions:

• Make predictions on new data and output the predicted failure types to a CSV file.

🏆 Results

• XGBoost Accuracy: 90.12% ✔️
• Random Forest Accuracy: 89.75% ✔️

The XGBoost model slightly outperforms the Random Forest model on the test dataset.

📂 File Structure

.
├── predictive_maintenance.csv          # Dataset for training
├── Sensor_Data.csv                     # Dataset on which predictions are made
├── xgb_model.pkl                       # Saved XGBoost model
├── rf_model.pkl                        # Saved Random Forest model
├── Sensor_Data_predictions.csv         # Dataset with predicted failure types