Data_2

The Machine Failure Predictions Dataset (D_2) is a real-world dataset sourced from Kaggle, containing 10,000 records and 14 features pertinent to IIoT device performance and health status. The binary target feature, 'failure', indicates whether a device is functioning (0) or has failed (1). Predictor variables include telemetry readings and categorical features related to device operation and environment. Data preprocessing included aggregating features related to failure types and removing non-informative features such as Product ID. The initial dataset exhibited a significant class imbalance (3.39\% failure rate), which was rectified using the SMOTE algorithm, resulting in a balanced dataset. To further enhance modeling accuracy, 3,865 synthetic samples were generated through the Gretel.ai platform. D_2 is suitable for predictive modeling tasks in IIoT, allowing researchers to explore machine failure patterns and improve maintenance planning for industrial systems.