README.md

metadata

title: Predictive Maintenance Engine Classifier
emoji: 🔧
colorFrom: blue
colorTo: green
sdk: docker
app_port: 7860
pinned: false

Predictive Maintenance Engine Classifier

This Hugging Face Space serves a Random Forest model for binary engine-condition prediction.

Configuration

Name	Default	Purpose
APP_HOST	0.0.0.0	Host used by the Streamlit server.
APP_PORT	7860	Port exposed by the Docker container and Hugging Face Space.
HF_MODEL_REPO	saranka85/predictive-maintenance-random-forest	Hugging Face model repository containing model.joblib.
HF_MODEL_FILENAME	model.joblib	Model artifact filename inside the model repository.
HF_TOKEN	optional	Required only if the model repository or Space is private.
PREDICTION_LOG_DIR	/tmp/prediction_logs	Runtime directory where submitted inputs are appended as CSV.
ENGINE_CLASS_0_LABEL	Requires Maintenance	Display label for engine_condition = 0.
ENGINE_CLASS_1_LABEL	Operating Normally	Display label for engine_condition = 1.

The app accepts six raw engine sensor values, saves them into a DataFrame, creates the engineered training features, loads the model from the Hugging Face Model Hub, and returns a maintenance decision:

• Requires Maintenance
• Operating Normally

The raw engine_condition class is still shown for traceability.