2026-10-11T02:51:25Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33940158 2026-09-18T17:57:51Z category_24 category_734 category_811 category_106 portal_5 item_type_3 month_year_09_2026

<p>External rolling-origin validation on five NYC 311 categories. Cost and fill are means across 26 test origins. Improvement is first calculated within each origin relative to Poisson and then averaged, giving each origin equal weight; Cost/day is instead the mean of absolute origin costs. Origins with different cost scales can therefore yield a positive mean paired improvement even when the displayed overall mean cost is higher.</p> Saisai Hou (9584629) Yunzhi Zhu (3837982) Sen Zhang (498066) Ying Chen (9697) Evolutionary Biology Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Science Policy type shares persist independence remains adequate fill rate increased 4 %&# 8211 2 %&# 8211 190 observations ), visit counts provide dominate model choice joint sensitivity analysis external temporal test deployment stress test ratio perturbations identify parameter learning curves 0 %&# 8211 independent poisson models independent poisson counts substantially calibrated model rand analysis nyc analysis sectional held scarf moment robust decisions reproducible check parameter error often converted matching environment latent heterogeneity inventory decisions five categories decision calibration decision calibrated count simulation confidence interval composition dependence based decisions asks whether although clustering 8 %. 096 days <p>External rolling-origin validation on five NYC 311 categories. Cost and fill are means across 26 test origins. Improvement is first calculated within each origin relative to Poisson and then averaged, giving each origin equal weight; Cost/day is instead the mean of absolute origin costs. Origins with different cost scales can therefore yield a positive mean paired improvement even when the displayed overall mean cost is higher.</p> 2026-09-18T17:57:38Z Dataset Dataset 10.1371/journal.pone.0358767.t003 https://figshare.com/articles/dataset/_p_External_rolling-origin_validation_on_five_NYC_311_categories_Cost_and_fill_are_means_across_26_test_origins_Improvement_is_first_calculated_within_each_origin_relative_to_Poisson_and_then_averaged_giving_each_origin_equal_weight_Cost_da/33940158 CC BY 4.0