Achievement Award, period of 10/01/2023 to 01/25/2024
Luca Sartore
Luca Sartore proposes a novel approach to generate a synthetic multidimensional dataset while preserving implicit collinearity among variables. The proposed algorithm can efficiently generate synthetic records. He also developed new methods to calculate global utility and risk. He also displays a helpful, cooperative, and positive attitude towards coworkers. Despite the work he must do, he consistently finds ways to assist his coworkers in completing RDD projects. For SDL project, his innovative techniques to generate synthetic data and to evaluate global utility, developed in a remarkably short time frame, have immensely benefited the synthetic sub team and also the SDL team. In addition, he also provided technical expertise to the on-going research effort to identify critical timing and climate indicators to monitoring winter wheat. For SDL project, based on the new approach for synthetic data and global utility measures, two abstracts are drafted and ready for the conferences after reviews. Both of them are expected to be papers and sent to peer review journal. In addition, his contributions on the winter wheat project were invaluable to completing deadlines and milestones for the 2024 growing season.
Lu Chen
As the survey modeling sub team lead, Lu worked with team members to implement spatio-temporal models at both the state and district level, ensuring their successful integration into NASS’s data and computation environment. She identified areas for improvement and devised strategies to outperform the initial models. Through effective leadership and collaboration, Lu guided the team in implementing these enhancements. The the model performances were also compared internally and externally. She presented the results to the IMAGES team and also in the RDD seminar, demonstrating the comparisons on both state-level and district-level spatio-temporal models.
She presented the results to the IMAGES team and also in the RDD seminar, demonstrating the comparisons on both state-level and district-level spatio-temporal models. She provided the recommendation to use the state-level model. The model could provide generally better results than the survey and even June board numbers. In addition, she actively worked and collaborated with colleagues within RDD and also from MD and SD to get insights and data for the district-level model.
Ruiyi Zhang
Ruiyi contributed to the Statistical Disclosure project by developing the audit prototype to evaluate the under-suppression by attacking the suppressed tables using only the margin total/linkage relations. Ruiyi worked on the development of the Cplex prototype to solve the Mixed-Integer-Programming model for 2d tables. Ruiyi is a team member of an IDEAL subteam to test Jimmy. Jimmy is a user interface developed to manage/share edit logic and track status of values. Ruiyi is part of the Inflation Reduction Act (IRA) Greenhouse Gas (GHG) Quantification Action Area #6 subteam, where the goal is to combat climate change to support America’s working Lands, natural resources and communities. Ruiyi joined the Sample Review Team with the purpose of NORC reviewed the sampling methodology for 83 of NASS’s surveys.
Ruiyi, as a team member of the Statistical Disclosure Limitation project, proposed a new cost in the objective function of the Mixed-Integer-Programming model for 2d tables, which can result in better data utility than minimizing pure complementary cell number of value. Ruiyi also evaluated two ways to handle 3d/linked/Hierarchical tables as follows:
a) A big Mixed-Integer-Programming model (Census Bureau/EIA) and
b) Break up into a series 2d tables and use backtracking (Current NASS/Tau-Argus Modular Algorithm).
Ruiyi exceeded expectations by completing test runs of Jimmy with crops APS data to find out its limitation and improve it. Ruiyi exceeded expectations by working as part of an inter-agency team with ERS, NRCS and ARS scientists to improve temporal and spatial coverage of national conservation activity data. Ruiyi efforts contributed to the research and implementation to reduce the CVs, cost, and respondent burden. As part of the Statistical Disclosure Limitation project, he result of Ruiyi's efforts determined that the better option is to use the Current NASS / Tau-Argus modular algorithm because of computational efficiency. Ruiyi's diligence is moving forward the research and evaluation of statistical disclosure methods. Several test runs of the Jimmy were completed, which were analyzed along with discussion of the results. As part of the team, Ruiyi contributed to the development of the current work plan, which now moves forward to an approval process through NRCS and OCE leadership. The result of Ruiyi's team work allows the Sample Review team to focus on the solutions that will have the most impact and can be implemented in 2024 or 2025 growing season.