Solar Flare Detection Dataset from a Dual-Mode Triggering and Nested Event Segmentation Algorithm Based on GOES 1–8 Å X-ray Flux

Data processing is completed using Python (NumPy, Pandas). The time span of the dataset corresponds to the maximum period of the 25th solar activity cycle, and is strictly divided into a calibration set (May 2024) and an independent test set (the entire year of 2025) according to time segmentation. The calibration set directory calibration_202405 contains three files: goes_xray_202405_1min.txt for GOES X-ray flux (44618 records, listed as timestamp, 1-8 Å flux, 0.5-4 Å flux, in W/m ²); Swpc_flare_list_202405. csv is the official SWPC flare list (378 records, including start and end time, peak time, flare classification, and other fields); AlgorithmDetection_202405. csv is the algorithm detection result (485 records, including start and end time, peak flux, and flare classification).

The test set directory test_2025 contains five files: goes_xray_2025_full_1min.txt represents the annual GOES flux (525186 records); Swpc_flare_list_2025_full. csv is the annual SWPC list (3294 entries, C level 2793, M level 341, X level 18, B level 142); AlgorithmDetection_2025_full. csv is the annual algorithm detection result (3899 items); Summary_overall_tetrics.csv is a summary of the annual performance levels (C/M/X/Overall, including precision, recall, etc.) F1）； Summary'monthly_metricsv is a monthly grading index (36 items, 12 months, divided into C/M/X levels for each month).

All file formats are CSV or TXT, and the data can be used for the replication and validation of flare detection algorithms, performance comparison, and research on commercial monitoring of space weather.