Predictive Maintenance for Server Metrics

This workspace contains a dataset and a production-ready Python pipeline for predictive maintenance using server monitoring time series.

Files

• system-1.csv — downloaded server metrics dataset from the public westermo/test-system-performance-dataset repository.
• pdm_server_maintenance.py — modular pipeline for feature engineering, normalization, time-series sequence creation, LSTM/GRU modeling, Isolation Forest anomaly detection, and visualization.
• requirements.txt — required Python packages.

Usage

1. Install dependencies:

pip install -r requirements.txt

2. Run the pipeline:

python pdm_server_maintenance.py --dataset system-1.csv --window-size 20 --model-type gru --epochs 30

3. If the dataset is missing, add --download to fetch it automatically:

python pdm_server_maintenance.py --download

Output

• anomaly_timeline.png — generated plot of anomaly score and server status over time.