Skip to content

README.md

Latest commit

 

History

History
59 lines (42 loc) · 1.7 KB

README.md

File metadata and controls

59 lines (42 loc) · 1.7 KB

🔧 Predictive Maintenance: Failure Classification

This project demonstrates predictive maintenance by classifying machine failures

The project leverages a synthetic dataset designed to mimic real-world predictive maintenance scenarios.

Google Colab Link

Open in Colab

🌟 Introduction

Predictive maintenance is a proactive approach that utilizes data analysis to predict when equipment failure might occur. By anticipating failures, timely maintenance can be performed, reducing downtime and saving costs.

📊 Dataset Overview

The synthetic dataset simulates real-world maintenance scenarios with:

  • 10,000 records and 14 features
  • Target variable:
    • Failure_Type: Specifies the type of failure (multiclass labels).
  • Features include:
    • Continuous: Temperature, rotational speed, torque, tool wear, etc.
    • Categorical: Product quality, serial numbers, etc.

🚀 Project Workflow

  1. EDA & Data Preperations:

    • Statistical Analysis
    • Visualizations
    • Handling missing values and outliers
    • Skewness Analysis
    • Correlation Analysis

  2. Feature Engineering & Normalization:

    • Three new features were engineered
    • Ordinal and Standard normalization

  3. Model Training:

    • Multiclass classification using algorithms like Decision Trees, Gradient Boosting, etc.

  4. Evaluation:

    • Metrics: Accuracy, Precision, Recall, F1-Score, Support
    • Classification report

🛠️ Requirements

  • Python 3.8 or above
  • Libraries:
    • pandas
    • numpy
    • scikit-learn
    • matplotlib