Global 12-Month Lunar Crescent Visibility Dataset for Machine Learning Prediction

OverviewThis dataset supports the research presented in the article “Toward a Globally Lunar Calendar: A Machine Learning-Driven Approach for Crescent Moon Visibility Prediction,” published in the Journal of Big Data in 2024. The dataset was developed to support machine-learning-based prediction of new crescent Moon visibility across all twelve lunar months and multiple geographical regions. It extends earlier Ramadan-focused research toward crescent visibility prediction throughout the complete lunar year. Dataset Construction and ProvenanceThe dataset was independently compiled and structured by the authors from publicly accessible crescent observation records available through the International Crescent Observation Project (ICOP). At the time of data collection, the relevant observation information was distributed across multiple ICOP webpages, countries, observation reports, and individual records rather than being available as a single structured machine-readable dataset. The authors manually collected the relevant factual observation information and subsequently organized, translated, standardized, cleaned, and transformed it into a structured research dataset. Additional astronomical variables were calculated or obtained using Accurate Times 5.6 software, as described in the associated publication. The original compiled dataset contains 2,085 observations from 47 countries, covering all twelve lunar months over a 13-year period from 2010 (1431 Hijri) to 2023 (1444 Hijri). Following the preprocessing procedures described in the associated publication, 1,779 observations remained for machine-learning analysis. VariablesThe dataset includes Hijri Day, Conjunction Time, Date, Country, City, State, Atmosphere, V_eye, V_bino, V_tele, V_ccd, Longitude, Latitude, Sunset, Moonset, Sun_Moon_Lag, Age_of_the_Moon, Moon_altitude, Sun_altitude, Altitude_difference, Moon_azimuth, Sun_azimuth, Azimuth_difference, Elongation, and Illumination. The V_eye variable represents naked-eye visibility of the new crescent Moon and served as the principal prediction target in the associated machine-learning study. Source AcknowledgementCrescent observation facts used in constructing this dataset were collected from publicly accessible observation records of the International Crescent Observation Project (ICOP). ICOP is acknowledged as the source of the original crescent observation reports. The structured dataset, data integration, organization, translation, standardization, preprocessing, and derived variables were produced by the dataset authors. No ICOP photographs, website layouts, or other copyrighted third-party materials are included in this dataset. Associated PublicationLoucif, S., Al-Rajab, M., Abu Zitar, R., & Rezk, M. (2024). Toward a globally lunar calendar: a machine learning-driven approach for crescent moon visibility prediction. Journal of Big Data, 11, Article 114. DOI: 10.1186/s40537-024-00979-6. Access and Conditions of UseThe dataset files are provided under restricted access. Researchers may request access for legitimate academic and research purposes. Applicants are required to provide information concerning their institutional affiliation, proposed research, intended use of the dataset, additional researchers who will access the data, and anticipated research outputs, and to agree to the applicable Data Use Agreement and citation requirements. Researchers granted access are required to appropriately cite the Zenodo dataset DOI and the associated publication in scholarly outputs substantially based on the dataset. ICOP should also be acknowledged as the source of the original crescent observation reports where appropriate. Redistribution, public reposting, or unauthorized transfer of the restricted dataset is not permitted under the conditions of access.