Plot log loss

pdr_backend/sim/dash_plots/view_elements.py

@@ -14,6 +14,7 @@
     "aimodel_varimps",
     "aimodel_response",
     "f1_precision_recall_vs_time",
+    "log_loss_vs_time",
 ]
 
 empty_slider = dcc.Slider(
@@ -59,49 +60,27 @@ def get_header_elements(run_id, st, ts):
     ]
 
 
+def side_by_side_graphs(figures, name1, name2):
+    return html.Div(
+        [
+            dcc.Graph(figure=figures[name1], id=name1, style={"width": "50%"}),
+            dcc.Graph(figure=figures[name2], id=name2, style={"width": "50%"}),
+        ],
+        style={"display": "flex", "justifyContent": "space-between"},
+    )
+
+
 def arrange_figures(figures):
     return [
+        side_by_side_graphs(figures, "pdr_profit_vs_time", "trader_profit_vs_time"),
         html.Div(
             [
-                dcc.Graph(figure=figures["pdr_profit_vs_time"], style={"width": "50%"}),
-                dcc.Graph(
-                    figure=figures["trader_profit_vs_time"], style={"width": "50%"}
-                ),
-            ],
-            style={"display": "flex", "justifyContent": "space-between"},
-        ),
-        html.Div(
-            [
-                dcc.Graph(figure=figures["accuracy_vs_time"]),
-            ]
-        ),
-        html.Div(
-            [
-                dcc.Graph(
-                    figure=figures["pdr_profit_vs_ptrue"], style={"width": "50%"}
-                ),
-                dcc.Graph(
-                    figure=figures["trader_profit_vs_ptrue"], style={"width": "50%"}
-                ),
-            ],
-            style={"display": "flex", "justifyContent": "space-between"},
-        ),
-        html.Div(
-            [
-                dcc.Graph(
-                    figure=figures["aimodel_varimps"],
-                    id="aimodel_varimps",
-                    style={"width": "50%"},
-                ),
-                dcc.Graph(figure=figures["aimodel_response"], style={"width": "50%"}),
-            ],
-            style={"display": "flex", "justifyContent": "space-between"},
-        ),
-        html.Div(
-            [
-                dcc.Graph(figure=figures["f1_precision_recall_vs_time"]),
+                dcc.Graph(figure=figures["accuracy_vs_time"], id="accuracy_vs_time"),
             ]
         ),
+        side_by_side_graphs(figures, "pdr_profit_vs_ptrue", "trader_profit_vs_ptrue"),
+        side_by_side_graphs(figures, "aimodel_varimps", "aimodel_response"),
+        side_by_side_graphs(figures, "f1_precision_recall_vs_time", "log_loss_vs_time"),
     ]
 
 

pdr_backend/sim/sim_plotter.py

@@ -311,6 +311,25 @@ def plot_trader_profit_vs_ptrue(self):
 
         return fig
 
+    @enforce_types
+    def plot_log_loss_vs_time(self):
+        clm = self.st.clm
+        s = f"log loss = {clm.losses[-1]:.4f}"
+
+        y = "log loss"
+        df = pd.DataFrame(clm.losses, columns=[y])
+        df["time"] = range(len(clm.losses))
+
+        fig = go.Figure(
+            go.Scatter(x=df["time"], y=df[y], mode="lines", name="log loss")
+        )
+
+        fig.update_layout(title=s)
+        fig.update_xaxes(title="time")
+        fig.update_yaxes(title=y)
+
+        return fig
+
 
 def file_age_in_seconds(pathname):
     stat_result = os.stat(pathname)

pdr_backend/sim/test/test_dash_plots.py

@@ -39,16 +39,12 @@ def test_arrange_figures():
     figures = {key: Figure() for key in figure_names}
     result = arrange_figures(figures)
     seen = set()
-    count = 0
     for div in result:
         for graph in div.children:
-            count += 1
             if hasattr(graph, "id"):
                 seen.add(graph.id)
 
-    assert count == len(figure_names)
-    assert seen == {"aimodel_varimps"}  # only one with id present for now
-    set(figure_names)
+    assert len(seen) == len(figure_names)
 
 
 def test_snapshot_slider():
@@ -91,8 +87,11 @@ def test_get_figures_by_state():
     mock_sim_plotter.plot_pdr_profit_vs_ptrue.return_value = Figure()
     mock_sim_plotter.plot_trader_profit_vs_ptrue.return_value = Figure()
     mock_sim_plotter.plot_f1_precision_recall_vs_time.return_value = Figure()
+    mock_sim_plotter.plot_log_loss_vs_time.return_value = Figure()
+
     plotdata = Mock()
     plotdata.colnames = ["var1", "var2"]
+
     mock_sim_plotter.aimodel_plotdata = plotdata
 
     with patch(