JHUAPL DMD metadata (#4)

Johns Hopkins University Applied Physics Labs Dynamic Mode Decomposition model metadata

Co-authored-by: William Redman <[email protected]>

model-metadata/JHUAPL-DMD.yml

@@ -0,0 +1,26 @@
+team_name: "Johns Hopkins University Applied Physics Lab"
+team_abbr: "JHUAPL"
+model_name: "Dynamic mode decomposition"
+model_abbr: "DMD"
+model_version: "1.0"
+model_contributors: [
+  {
+    "name": "William T. Redman",
+    "affiliation": "Johns Hopkins University Applied Physics Lab",
+    "email": "[email protected]"
+  },
+  {
+    "name": "Luke Mullany",
+    "affiliation": "Johns Hopkins University Applied Physics Lab",
+    "email": "[email protected]"
+  }
+]
+website_url: "https://www.jhuapl.edu/"
+license: "CC-BY-4.0"
+team_funding: "ACCIDDA"
+designated_model: true
+methods: "Dynamic mode decomposition for time-series."
+data_inputs: "Daily and weekly incident flu hospitalizations, queried through FluSight"
+methods_long: "A sliding window is used to construct an temporally local approximation of the Koopman operator. Time-delays and radial basis functions are used for the lifting to function space. All regions, provinces, and national data are used to construct the approximation of the Koopman operator to estimate couplings/correlations. Gaussian noise, of variance is used to sample possible variants of the data to approximate the forecast percentiles."
+ensemble_of_models: false
+ensemble_of_hub_models: false