add data-sort-value attr to metrics table cells in "model params" and

"training size" cols + bump [email protected]

closes Improper sorting on "Model Params" #124

scripts/model_figs/metrics_tables.py

@@ -9,7 +9,7 @@
 
 from matbench_discovery import PDF_FIGS, SCRIPTS, SITE_FIGS
 from matbench_discovery.data import DataFiles, df_wbm
-from matbench_discovery.enums import MbdKey, ModelType, Open
+from matbench_discovery.enums import MbdKey, Open
 from matbench_discovery.metrics import stable_metrics
 from matbench_discovery.models import MODEL_METADATA
 from matbench_discovery.preds import df_metrics, df_metrics_10k, df_metrics_uniq_protos
@@ -60,12 +60,16 @@
         n_structs = train_set_meta["n_structures"]
         n_materials = train_set_meta.get("n_materials", n_structs)
         title = "Number of materials in training set"
-        train_size_str = f"<span {title=}>{si_fmt(n_materials, fmt='.0f')}</span>"
+        train_size_str = (
+            f"<span {title=} data-sort-value={n_materials}>"
+            f"{si_fmt(n_materials, fmt='.0f')}</span>"
+        )
 
         if n_materials != n_structs:
             title = "Number of materials (and structures) in training set"
             train_size_str = (
-                f"<span {title=}>{si_fmt(n_materials, fmt='.1f')}</span> "
+                f"<span {title=} data-sort-value={n_materials}>"
+                f"{si_fmt(n_materials, fmt='.0f')}</span>"
                 f"<small {title=}>({si_fmt(n_structs, fmt='.1f')})</small>"
             )
 
@@ -76,23 +80,20 @@
             )
 
         df_met.loc[Key.train_set.label, model] = train_size_str
-    model_params = model_data.get(Key.model_params, "")
+    model_params = model_data.get(Key.model_params, 0)
     n_estimators = model_data.get(Key.n_estimators, -1)
     title = "Number of models in ensemble"
     n_estimators_str = (
         f" <small {title=}>(N={n_estimators})</small>" if n_estimators > 1 else ""
     )
 
-    model_type = model_data.get(Key.model_type, "")
-    title = ModelType.val_label_dict().get(model_type) or ""
-    df_met.loc[Key.model_type.label, model] = f"<span {title=}>{model_type}</span>"
-
     title = "Number of trainable model parameters"
+    formatted_params = si_fmt(model_params)
     df_met.loc[Key.model_params.label.replace("eter", ""), model] = (
-        f"<span {title=}>{si_fmt(model_params)}</span>{n_estimators_str}"
-        if isinstance(model_params, int)
-        else model_params
+        f'<span {title=} data-sort-value="{model_params}">{formatted_params}'
+        f"</span>{n_estimators_str}"
     )
+
     for key in (
         MbdKey.openness,
         Key.train_task,
@@ -194,10 +195,10 @@
             na_rep="",  # render NaNs as empty string
         )
         .background_gradient(
-            cmap="viridis", subset=list(higher_is_better & {*df_table}), axis=0
+            cmap="viridis", subset=list(higher_is_better & {*df_table}), axis="index"
         )
         .background_gradient(  # reverse color map if lower=better
-            cmap="viridis_r", subset=list(lower_is_better & {*df_table}), axis=0
+            cmap="viridis_r", subset=list(lower_is_better & {*df_table}), axis="index"
         )
     )
     # add up/down arrows to indicate which metrics are better when higher/lower

scripts/wbm_umap_projection.py

@@ -156,8 +156,8 @@ def features_to_drop(df_in: pd.DataFrame, threshold: float = 0.95) -> list[str]:
     df_wbm = pd.read_csv(wbm_matminer_feat_path).set_index(Key.mat_id)
 
     # Drop all rows containing NaN values
-    df_wbm = df_wbm.dropna(axis=0)
-    df_mp = df_mp.dropna(axis=0)
+    df_wbm = df_wbm.dropna(axis="index")
+    df_mp = df_mp.dropna(axis="index")
 
     # Drop highly correlated features
     cols_to_drop = features_to_drop(df_mp, threshold=0.95)

site/eslint.config.js

@@ -11,7 +11,7 @@ export default [
         `error`,
         { argsIgnorePattern: `^_`, varsIgnorePattern: `^_` },
       ],
-      '@typescript-eslint/quotes': [`error`, `backtick`, { avoidEscape: true }],
+      '@/quotes': [`error`, `backtick`, { avoidEscape: true }],
       'svelte/no-at-html-tags': `off`,
     },
   },

site/package.json

@@ -19,19 +19,19 @@
   "devDependencies": {
     "@iconify/svelte": "^4.0.2",
     "@rollup/plugin-yaml": "^4.1.2",
-    "@sveltejs/adapter-static": "^3.0.2",
-    "@sveltejs/kit": "^2.5.20",
+    "@sveltejs/adapter-static": "^3.0.4",
+    "@sveltejs/kit": "^2.5.24",
     "@sveltejs/vite-plugin-svelte": "^3.1.1",
     "d3-scale-chromatic": "^3.1.0",
     "elementari": "^0.2.3",
-    "eslint": "^9.8.0",
+    "eslint": "^9.9.0",
     "eslint-plugin-svelte": "^2.43.0",
     "hastscript": "^9.0.0",
     "highlight.js": "^11.10.0",
     "js-yaml": "^4.1.0",
     "json-schema-to-typescript": "^15.0.0",
     "katex": "^0.16.11",
-    "mdsvex": "^0.11.2",
+    "mdsvex": "^0.12.3",
     "prettier": "^3.3.3",
     "prettier-plugin-svelte": "^3.2.6",
     "rehype-autolink-headings": "^7.1.0",
@@ -43,12 +43,12 @@
     "svelte-multiselect": "^10.3.0",
     "svelte-preprocess": "^6.0.2",
     "svelte-toc": "^0.5.9",
-    "svelte-zoo": "^0.4.10",
+    "svelte-zoo": "^0.4.11",
     "svelte2tsx": "^0.7.15",
     "tslib": "^2.6.3",
     "typescript": "5.5.4",
-    "typescript-eslint": "^8.0.1",
-    "vite": "^5.3.5"
+    "typescript-eslint": "^8.1.0",
+    "vite": "^5.4.1"
   },
   "prettier": {
     "semi": false,