Simplify filters to separate dataset and metric checkboxes

app.py

@@ -139,30 +139,23 @@ def build_tab(df, name):
                     elem_classes="filter-group"
                 )
 
-                # Add metric filters
-                gr.Markdown("**Filter by Metrics**")
-                metric_filters = {}
-                for metric in metric_cols:
-                    # Get all numeric values for this metric across all dataset columns
-                    metric_values = []
-                    for col in metric_combo_cols:
-                        if f"({metric})" in col:
-                            values = pd.to_numeric(pivoted_df[col], errors='coerce').dropna()
-                            metric_values.extend(values.tolist())
-
-                    if metric_values:
-                        min_val = min(metric_values)
-                        max_val = max(metric_values)
-                        metric_filters[metric] = gr.Slider(
-                            minimum=min_val,
-                            maximum=max_val,
-                            value=min_val,
-                            step=(max_val - min_val) / 100,
-                            label=f"Min {metric}",
-                            elem_classes="metric-filter"
-                        )
-
-        def update(search, md, org, cols, *metric_values):
+                gr.Markdown("**Datasets**")
+                dataset_filter = gr.CheckboxGroup(
+                    choices=datasets,
+                    value=datasets,
+                    label="",
+                    elem_classes="filter-group"
+                )
+
+                gr.Markdown("**Metrics**")
+                metric_filter = gr.CheckboxGroup(
+                    choices=metric_cols,
+                    value=metric_cols,
+                    label="",
+                    elem_classes="filter-group"
+                )
+
+        def update(search, md, org, dset, metrics, cols):
             filtered = pivoted_df.copy()
 
             if search:
@@ -174,33 +167,33 @@ def build_tab(df, name):
             if org:
                 filtered = filtered[filtered["Organization"].isin(org)]
 
-            # Apply metric filters
-            metric_names = list(metric_filters.keys())
-            for i, metric_name in enumerate(metric_names):
-                min_threshold = metric_values[i]
-                # Filter rows where at least one column with this metric meets the threshold
-                metric_columns = [col for col in metric_combo_cols if f"({metric_name})" in col and col in filtered.columns]
-                if metric_columns:
-                    mask = filtered[metric_columns].apply(
-                        lambda row: any(pd.to_numeric(val, errors='coerce') >= min_threshold if pd.notna(val) and val != "-" else False for val in row),
-                        axis=1
-                    )
-                    filtered = filtered[mask]
-
+            # Filter by dataset and metric - hide columns that don't match
             if cols:
-                display_cols = [col for col in cols if col in filtered.columns]
+                display_cols = []
+                # Always include base columns
+                for col in ["Model", "Organization", "First Author"]:
+                    if col in cols and col in filtered.columns:
+                        display_cols.append(col)
+
+                # Add metric columns that match selected datasets and metrics
+                for col in cols:
+                    if col in metric_combo_cols:
+                        # Check if this column matches selected datasets and metrics
+                        col_dataset = col.split(" (")[0]
+                        col_metric = col.split(" (")[1].rstrip(")")
+                        if col_dataset in dset and col_metric in metrics:
+                            display_cols.append(col)
+
                 filtered = filtered[display_cols]
 
             return filtered
 
-        inputs = [search_bar, model_filter, org_filter, col_selector] + list(metric_filters.values())
-
-        search_bar.change(update, inputs, table)
-        model_filter.change(update, inputs, table)
-        org_filter.change(update, inputs, table)
-        col_selector.change(update, inputs, table)
-        for metric_filter in metric_filters.values():
-            metric_filter.change(update, inputs, table)
+        search_bar.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
+        model_filter.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
+        org_filter.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
+        dataset_filter.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
+        metric_filter.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
+        col_selector.change(update, [search_bar, model_filter, org_filter, dataset_filter, metric_filter, col_selector], table)
 
 
 custom_css = """
@@ -283,14 +276,6 @@ custom_css = """
     font-size: 14px;
 }
 
-.metric-filter {
-    margin-bottom: 1em;
-}
-
-.metric-filter label {
-    font-size: 14px;
-}
-
 .column-select {
     margin-bottom: 1.5em;
 }