Hadoop CNN-SPN Synthetic Dataset for Performance Monitoring and Reliability Analysis

This repository contains the synthetic dataset hadoop_cnn_spn_dataset_ROC.csv, developed to support the experimental evaluation of a hybrid Convolutional Neural Network (CNN) and Stochastic Petri Net (SPN) framework for Hadoop cluster performance monitoring and reliability analysis. The dataset was obtained through a collection of observations recorded from Hadoop cluster behavior under various workload scenarios. These observations integrate system-level performance indicators, SPN state variables, transition firing attributes, and probabilistic HFiringT scores computed from different firing-time distributions. The resulting dataset reflects the stochastic evolution of Hadoop execution processes and provides a reliable basis for evaluating machine learning and SPN-based approaches for performance analysis and reliability prediction. The dataset comprises 10,000 synthetic observations and includes features describing CPU utilization, memory usage, disk I/O, network bandwidth, task failures, container activity, heartbeat events, Petri net token markings, transition firing times, and normalized probability scores computed from Exponential, Normal, Log-Normal, Weibull, Erlang, and Combined distributions. This dataset enables researchers to reproduce the experimental results reported in the associated publication, including: Performance evaluation of the proposed CNN-SPN framework. Receiver Operating Characteristic (ROC) analysis. Confusion matrix generation. Statistical analysis of HFiringT probability distributions. Comparative evaluation of multiple probabilistic firing-time models. The repository promotes reproducible research by providing the dataset together with the Python scripts required to regenerate the reported figures and performance metrics. Dataset characteristics Dataset name: Hadoop CNN-SPN Synthetic Dataset Format: CSV (UTF-8) Number of instances: 10,000 Number of attributes: 33 (depending on the released version) Data type: Synthetic Application domain: Cloud Computing, Hadoop Performance Monitoring, Stochastic Petri Nets, Machine Learning, Reliability Analysis Keywords: Hadoop, Convolutional Neural Networks, Stochastic Petri Nets, Cloud Computing, Performance Monitoring, Reliability Analysis, ROC Analysis, HFiringT, Synthetic Dataset, Machine Learning.