adding parquets

app.py

@@ -62,8 +62,8 @@ def down_samp(embedding):
 
 
 def data_comparison(df):
-    selection = alt.selection_multi(fields=['cluster:O','label:O'])
-    color = alt.condition(alt.datum.slice == 'high-loss', alt.Color('cluster:O', scale = alt.Scale(domain=df.cluster.unique().tolist())), alt.value("lightgray"))
+    selection = alt.selection_multi(fields=['cluster:N','label:O'])
+    color = alt.condition(alt.datum.slice == 'high-loss', alt.Color('cluster:N', scale = alt.Scale(domain=df.cluster.unique().tolist())), alt.value("lightgray"))
     opacity = alt.condition(selection, alt.value(0.7), alt.value(0.25))
 
     # basic chart
@@ -72,7 +72,7 @@ def data_comparison(df):
         y=alt.Y('y:Q', axis=None),
         color=color,
         shape=alt.Shape('label:O', scale=alt.Scale(range=['circle', 'diamond'])),
-        tooltip=['cluster:O','slice:N','content:N','label:O','pred:O'],
+        tooltip=['cluster:N','slice:N','content:N','label:O','pred:O'],
         opacity=opacity
     ).properties(
         width=1000,
@@ -81,7 +81,7 @@ def data_comparison(df):
 
     legend = alt.Chart(df).mark_point(size=100, filled=True).encode(
         x=alt.X("label:O"),
-        y=alt.Y('cluster:O', axis=alt.Axis(orient='right'), title=""),
+        y=alt.Y('cluster:N', axis=alt.Axis(orient='right'), title=""),
         shape=alt.Shape('label:O', scale=alt.Scale(
         range=['circle', 'diamond']), legend=None),
         color=color,