Machine Failure Analysis with Python

Exploratory analysis of machine failure data using pandas and Matplotlib.

Objectives

• Load and inspect the AI4I 2020 Predictive Maintenance dataset.
• Calculate the overall machine failure rate and rates for each failure type.
• Compare average torque, tool wear, and temperature difference between samples with and without machine failure.
• Visualize failure rates by type.

Tools

• Python
• pandas
• Matplotlib

Analysis

Machine failure status is represented by:

• 0: No machine failure
• 1: Machine failure

Temperature difference is calculated for each sample as:

Process temperature [K] − Air temperature [K]

Average measurements are compared by machine failure status. The temperature difference chart includes all machine failures; it is not an analysis restricted to HDF.

Results

• Total samples: 10,000
• Samples with machine failure: 339
• Samples without machine failure: 9,661
• Overall machine failure rate: 3.39%
• No missing values were found.

Failure rates by type

Each rate is calculated relative to all 10,000 samples. A sample can have more than one failure type, so these percentages should not be added to calculate the overall machine failure rate.

Failure type	Rate (%)
TWF	0.46
HDF	1.15
PWF	0.95
OSF	0.98
RNF	0.19

Comparison of averages

Measurement	Without failure	With failure
Torque (Nm)	39.63	50.17
Tool wear (min)	106.69	143.78
Rotational speed (rpm)	1540.26	1496.49

Samples with failure had higher average torque and tool wear, and lower average rotational speed.

Average temperature differences are shown in the chart below and saved in comparison_summary.csv.

These comparisons describe associations and do not establish causation. This project performs exploratory analysis and does not train a prediction model.

Charts

Torque

Tool wear

Temperature difference

Failure rates by type

How to run

1. Install Python.
2. Download or clone this repository.
3. Ensure ai4i2020.csv is in the same folder as main.py.
4. Open a terminal in the project folder.
5. Install the required libraries:

python -m pip install -r requirements.txt

6. Run the analysis:

python main.py

Close each chart window to continue to the next chart. The script prints analysis results and saves CSV tables and PNG charts in the current working directory.

Output files

File	Contents
comparison_summary.csv	Average torque, tool wear, rotational speed, and temperature difference by failure status
failure_type_rates.csv	Percentage of all samples with each failure type
average_torque.png	Average torque by failure status
average_tool_wear.png	Average tool wear by failure status
average_temperature_difference.png	Average process minus air temperature by failure status
failure_type_rates.png	Failure rates by type

Dataset source

AI4I 2020 Predictive Maintenance Dataset (2020). UCI Machine Learning Repository.

• Dataset: https://archive.ics.uci.edu/dataset/601/ai4i
• DOI: https://doi.org/10.24432/C5HS5C
• License: CC BY 4.0
• License link: https://creativecommons.org/licenses/by/4.0/

The dataset contains 10,000 synthetic samples representing industrial predictive maintenance conditions.

The original CSV is used without modification. Temperature difference is calculated during analysis as an additional column.