Dataset for the paper "Review on Transit Method and Artificial Intelligence for Detecting Long-Period Weak Signals"

This dataset is aimed at the comparative study of Kepler exoplanet transit signal detection and photometric noise. It collects traceable input data of key charts in the paper and author derived result data, and is divided into three categories: the first is publicly available archive download data (Class A), which includes the planetary system comprehensive catalog (PS) and KOI cumulative candidate catalog (cumulative, both fixed with file name timestamps on March 4, 2026) obtained from NASA Exoplanet Archive, as well as several Kepler target star light curve metadata downloaded from MAST (A4-A9); The second type is derived data (Class B) generated by the author after preprocessing, noise modeling, and signal injection based on publicly available data, which is used to reproduce the noise characteristics, injection of stars, and detection performance in the paper's figures; The third is a summary of parameters used for complete reproduction (Class C). The data processing is mainly completed in the Python environment (README provides Python ≥ 3.9 and dependencies such as numpy, scientific, matplotlib, pandas optional, lightkurve, etc.), where the publicly available Kepler light curve can be retrieved and downloaded from MAST through lightkurve; Subsequently, a unified timeline was organized and simulated/injected into the light sequence (B1-B4), and results consistent with this dataset were exported under fixed random seeds; In the detection stage, BLS and TLS periodic searches were performed on the same target (KIC 9100953, Kepler-1610) to obtain two periodic plots (B5, B6), which were used to compare the changes in statistics under different probing periods and locate significant candidate periodic peaks. In terms of time information, directory files provide version traceability through download timestamps; The simulation/injection data adopts a relative time axis with "days" as the unit (consistent with the figure in the paper), and the periodic graph files (B5, B6) cover a tentative periodic grid of 32.002-57.998 days; In terms of spatial information, the data is indexed by the target stars within Kepler's field of view (identified by KIC numbers) and does not include pixel level spatial grids or geographic spatial resolution fields, therefore spatial resolution is not applicable. The table data is all in CSV format, with rows representing independent record points and columns representing physical/statistical fields. For example, the BLS/TLS cycle chart contains 8347 records each, and the two columns are "cycle (days)" and "power/detection statistics (algorithm output)"; Provide the main peak period and main peak power under the same periodic grid (listed in the README). In terms of missing and erroneous information, directory and metadata files may have null/missing fields due to missing measurements or quality markings in the original archive fields (which is an inherent situation of upstream archives); The Kepler light sequence itself may also experience discontinuous sampling due to data quality rejection and observation discontinuity (introducing window function effects for periodic detection); Although simulation and injection data can be reproduced through fixed random seeds, they still contain methodological error sources caused by noise model assumptions, periodic grid discretization, and differences in BLS/TLS statistical definitions (therefore, the "power" column in the README is clearly labeled as the algorithm output and it is recommended to maintain the same implementation during reproduction). In terms of file format, this dataset uses universal CSV, which can be directly read by common software such as Excel, WPS tables, LibreOffice Calc, and Python/R/Matlab