Merge pull request #410 from jamiepratt/codex/pair-frequency-logistic-v2

Publish pair-frequency model validation

Author: timothypratley

deps.edn

@@ -1,4 +1,4 @@
-{:paths ["src"]
+{:paths ["src" "resources"]
 
  :deps
  {org.clojure/clojure                         {:mvn/version "1.12.3"}
@@ -100,4 +100,11 @@
                                org.scicloj/kindly-advice {:local/root "../kindly-advice"}
                                org.scicloj/kindly-render {:local/root "../kindly-render"}}}
   :neil       {:project {:name io.github.timothypratley/clojurecivitas}}
-  :dev        {:jvm-opts ["--add-opens=java.base/java.nio=ALL-UNNAMED"]}}}
+  :dev        {:jvm-opts ["--add-opens=java.base/java.nio=ALL-UNNAMED"]}
+  :test-clj   {:extra-paths ["test"]
+               :main-opts ["-m" "language-learning.vocabulary-estimation.test-runner"]}
+  :test-cljs  {:extra-paths ["test"]
+               :extra-deps {org.clojure/clojurescript {:mvn/version "1.12.42"}}
+               :main-opts ["-m" "cljs.main"
+                           "-re" "node"
+                           "-m" "language-learning.vocabulary-estimation.cljs-test-runner"]}}}

dev/generate_pair_frequency_fixture.clj

@@ -0,0 +1,40 @@
+(ns generate-pair-frequency-fixture
+  (:require [clojure.java.io :as io]
+            [clojure.string :as str]))
+
+(def source-path
+  (or (first *command-line-args*)
+      "/Users/jamiep/Documents/subtlex/data/database-import/surface_form_lemma_pair_frequency_ranks.tsv"))
+
+(def target-path
+  (or (second *command-line-args*)
+      "resources/language_learning/vocabulary_estimation/pair_frequency_fixture_v1.tsv"))
+
+(def source-columns
+  ["surface_form_lemma_pair_frequency_rank_id"
+   "surface_form_id"
+   "lemma_id"
+   "pair_frequency_sn_sum"
+   "pair_frequency_sn_sum_rank"])
+
+(defn select-row [indexes line]
+  (let [fields (str/split line #"\t" -1)]
+    (str/join "\t" (map #(nth fields %) indexes))))
+
+(with-open [reader (io/reader source-path)]
+  (let [[header & rows] (line-seq reader)
+        columns (str/split header #"\t" -1)
+        indexes (mapv #(.indexOf columns %) source-columns)
+        selected (take 8000 rows)]
+    (when (some neg? indexes)
+      (throw (ex-info "Required source column missing"
+                      {:required source-columns :actual columns})))
+    (when-not (= 8000 (count selected))
+      (throw (ex-info "Source has fewer than 8,000 data rows" {})))
+    (io/make-parents target-path)
+    (spit target-path
+          (str (str/join "\t" source-columns) "\n"
+               (str/join "\n" (map #(select-row indexes %) selected))
+               "\n"))))
+
+(println target-path)

dev/verify_pair_frequency_logistic_v2.clj

@@ -0,0 +1,56 @@
+(ns verify-pair-frequency-logistic-v2
+  (:require [clojure.java.io :as io]
+            [clojure.pprint :as pprint]
+            [language-learning.vocabulary-estimation.pair-frequency-logistic-v2 :as v2]))
+
+(def fixture-resource
+  "language_learning/vocabulary_estimation/pair_frequency_fixture_v1.tsv")
+
+(def output-path
+  "resources/language_learning/vocabulary_estimation/pair_frequency_logistic_v2_grid_check.edn")
+
+(defn fixture []
+  (let [pairs (-> (io/resource fixture-resource)
+                  slurp
+                  v2/parse-fixture
+                  v2/validate-fixture)
+        {:keys [pairs] :as transformed} (v2/frequency-transform pairs)]
+    (assoc transformed :pairs
+           (mapv #(assoc %1 :pair-index %2) pairs (range)))))
+
+(defn check-grid []
+  (let [{:keys [pairs]} (fixture)
+        xs (mapv :x pairs)
+        selected (->> (v2/selection-schedule pairs 8 v2/default-seed)
+                      (take 6)
+                      (mapcat identity)
+                      (mapv :pair-index))
+        threshold (v2/threshold-for-total xs 1.5 4000.0)
+        observations (v2/simulate-responses
+                      xs selected {:threshold threshold :width 1.5}
+                      2026071303)
+        coarse-posterior (v2/posterior-grid xs observations)
+        coarse-summary (v2/posterior-predictive-summary
+                        xs observations coarse-posterior 20000 2026071304)
+        doubled (v2/doubled-grid v2/default-grid)
+        doubled-posterior (v2/posterior-grid xs observations doubled
+                                             v2/default-prior)
+        doubled-summary (v2/posterior-predictive-summary
+                         xs observations doubled-posterior 20000 2026071304)
+        result {:fixture-id v2/fixture-id
+                :fixture-sha256 v2/fixture-sha256
+                :observations (mapv #(select-keys % [:pair-index :response])
+                                    observations)
+                :default-grid v2/default-grid
+                :doubled-grid doubled
+                :posterior-draws 20000
+                :seed 2026071304
+                :coarse coarse-summary
+                :doubled doubled-summary
+                :convergence (v2/grid-convergence coarse-summary
+                                                  doubled-summary)}]
+    (io/make-parents output-path)
+    (spit output-path (with-out-str (pprint/pprint result)))
+    result))
+
+(prn (select-keys (check-grid) [:coarse :doubled :convergence]))