c

A comprehensive analysis of factory sensor data for predictive maintenance and failure prediction.

📋 Overview

This project analyzes factory sensor data from 500,000 industrial machines to predict machine failures within 7 days. The analysis includes data exploration, visualization, statistical analysis, and machine learning models for failure prediction.

📁 Files

• data.ipynb - Main Jupyter notebook with complete analysis
• factory_sensor_simulator_2040.csv - Dataset containing 500,000 machine records: https://www.kaggle.com/datasets/canozensoy/industrial-iot-dataset-synthetic

🔍 Dataset Description

The dataset contains sensor readings and operational metrics for various industrial machines:

Machine Types (33 types)

• AGV, Shuttle System, Crane, Labeler, CNC Lathe, Industrial Chiller
• CMM, Pump, Vacuum Packer, Injection Molder, Automated Screwdriver
• Mixer, Valve Controller, Vision System, Laser Cutter, Compressor
• And 18 more types...

Features

• Machine Info:Machine_ID, Machine_Type, Installation_Year
• Sensor Data: Temperature, Vibration, Sound, Oil Level, Coolant Level, Power Consumption
• Maintenance: Last_Maintenance_Days_Ago, Maintenance_History_Count, Failure_History_Count
• AI Features: AI_Supervision, Error_Codes_Last_30_Days, AI_Override_Events
• Predictive: Remaining_Useful_Life_days, Failure_Within_7_Days (target)

Statistics

• Total Records: 500,000 machines
• Failure Rate: 6.01% (30,032 machines will fail within 7 days)
• Missing Data: Some specialty sensors (Laser_Intensity, Hydraulic_Pressure_bar, etc.) have high missingness rates

Project Goals

1. Exploratory Data Analysis (EDA)

   • Distribution of machine types
   • Failure rate analysis by machine type
   • Sensor reading comparisons between failed and healthy machines

2. Predictive Modeling

   • Build machine learning models to predict failures
   • Identify key indicators of impending failure
   • Evaluate model performance

3. Key Insights

   • Which sensor readings are most indicative of failure
   • Impact of AI supervision on failure rates
   • Maintenance schedule optimization opportunities