Final Technical Report - CGIAR NIR Database

cg.contactM.Sanchez-Garcia@cgiar.orgen_US
cg.contributor.centerInternational Center for Agricultural Research in the Dry Areas - ICARDAen_US
cg.contributor.centerInternational Maize and Wheat Improvement Center - CIMMYTen_US
cg.contributor.crpCGIAR Research Program on Wheat - WHEATen_US
cg.contributor.funderInternational Maize and Wheat Improvement Center - CIMMYTen_US
cg.contributor.projectCGIAR NIR Databaseen_US
cg.contributor.project-lead-instituteInternational Center for Agricultural Research in the Dry Areas - ICARDAen_US
cg.coverage.countryMAen_US
cg.coverage.regionNorthern Africaen_US
cg.creator.idSanchez-Garcia, Miguel: 0000-0002-9257-4583en_US
cg.date.embargo-end-dateTimelessen_US
cg.subject.agrovocwheaten_US
dc.contributorItria Ibba, Mariaen_US
dc.creatorSanchez-Garcia, Miguelen_US
dc.date.accessioned2021-11-10T13:40:39Z
dc.date.available2021-11-10T13:40:39Z
dc.description.abstractThe CGIAR NIR Database project aimed at providing CIMMYT and ICARDA (and eventually the OneCGIAR) with new tools for nutritional and end-use quality data management and to increase the number of traits that can be analyzed and the accuracy of these analysis using a non-destructive and cost-effective method such as NIRS. For it, a new on-line database was developed to standardize, store and manage wet chemistry and near-infrared spectroscopy multi-crop end-use quality data from CIMMYT and ICARDA quality laboratories. This database is now hosting more than the 26,272 samples of NIR spectra and 15,346 samples of wet chemistry quality data from both centers and will be continuously updated with the new samples analyzed every year from both laboratories. This standardized database, together with the API that makes the data available to be easily extracted in the desired format, opened the door to the use of BigData to create new and more precise models to predict destructive, time-consuming and expensive wet chemistry tests using the NIR spectra of grain, flour and straw samples. For it, a new analytical pipeline to develop and fine-tune new prediction models was created involving twelve pre-processing algorithms, three statistical models and machine learning regression algorithms. The cross-validation tests performed to confirm the capacity of the analytical pipeline to obtain accurate and predictive models showed accuracies of up to 98% for important agronomic and end-use quality traits such as grain protein content. This pipeline is being assembled in an R package that will be available to the international community.en_US
dc.formatPDFen_US
dc.identifierhttps://mel.cgiar.org/dspace/limiteden_US
dc.identifier.citationMiguel Sanchez-Garcia, Maria Itria Ibba. (1/11/2021). Final Technical Report - CGIAR NIR Database. Beirut, Lebanon: International Center for Agricultural Research in the Dry Areas (ICARDA).en_US
dc.identifier.statusTimeless limited accessen_US
dc.identifier.urihttps://hdl.handle.net/20.500.11766/66345
dc.languageenen_US
dc.publisherInternational Center for Agricultural Research in the Dry Areas (ICARDA)en_US
dc.rightsCopyrighted; all rights reserveden_US
dc.subjecttechnical reporten_US
dc.titleFinal Technical Report - CGIAR NIR Databaseen_US
dc.typeDonor Reporten_US
dcterms.available2021-11-01en_US

Files