A set of statistical data of gas explosion accident analysis analyzed by the research group
The sample of the accident analysis is a total of 194 heavy and extremely large gas explosion accidents that occurred in China from 2000 to 2009. The accident analysis did not use any accident cause theory, nor did it use classification methods for hierarchical division, so the cause relationship obtained by the analysis is flat. The causes of each accident are recorded separately. The data has undergone two corrections. The correction work has verified the reliability of the analysis, and in order to streamline the accident causes, similar causes have been merged. The number of accident causes obtained from the three sets of accident analysis (including two sets of corrections) are 379 (dataset origin.xls), 317 (dataset_after 1 correction.xls), and 192 (dataset_after 2 corrections.xls) respectively. The data used for visualization are the number of accident cases (horizontal axis) and the total number of accident causes obtained by cumulative analysis (vertical axis). Every time an accident is read, the reasons obtained from the accident analysis are added to the accident causes database. The total number of accidents causes in the current database is calculated and returned after a deduplication operation. Considering the small number of samples, referring to the idea of k-fold cross-validation, random sampling is performed when reading accidents. The visualization is done by python, and the algorithm is attached to the .py file bellow.