Dataset for Machine Learning-Based Event Classification and Vertex Reconstruction in MATE-TPC Nuclear Reaction Experiments

Each row represents the track coordinates of a particle, and each column corresponds to a specific physical quantity, such as spatial coordinates (x, y, z) and deposited charge (q). All variable definitions and units are consistent with those used in the main text of the paper.During data processing, conventional methods were first applied to label events and reconstruct the reaction vertex, while background events and non-physical events were removed.

Subsequently, machine learning models were trained using simulated data and then applied to experimental data to evaluate their performance. Due to the applied selection criteria, missing data are minimal, and events that do not satisfy the selection conditions are not included in the final dataset.Uncertainties in the dataset mainly arise from detector resolution, the accuracy of conventional data analysis methods, and the domain gap between experimental data and simulated data.

These factors may introduce systematic deviations in the results obtained from machine learning models.The dataset is stored in widely accessible formats and can be readily read and processed using common scientific computing tools, such as Python (NumPy, pandas) and ROOT.