changed colours of plots

changed colours of plots

bctools/plots.py

@@ -15,7 +15,7 @@
 
 from .thresholds import get_optimized_thresholds_df
 
-def predicted_proba_violin_plot(true_y, predicted_proba, threshold_step = 0.01, title = "Interactive Probabilities Violin Plot"):
+def predicted_proba_violin_plot(true_y, predicted_proba, threshold_step = 0.01, marker_size = 3, title = "Interactive Probabilities Violin Plot"):
 
     """
     Plots interactive and customized violin plots of predicted probabilties with plotly, 
@@ -37,6 +37,8 @@ def predicted_proba_violin_plot(true_y, predicted_proba, threshold_step = 0.01,
     threshold_step: float, default=0.01
         step between each classification threshold (ranging from 0 to 1) below which prediction label is 0, 1 otherwise
         each value will have a corresponding slider step
+    marker_size: int, default=3
+        Size of the points to be plotted
     title: str, default='Interactive probabilities Violin Plot'
         The main title of the plot.
     """
@@ -55,7 +57,7 @@ def predicted_proba_violin_plot(true_y, predicted_proba, threshold_step = 0.01,
     main_title = f"<b>{title}</b><br>"
 
     # VIOLIN PLOT 
-    full_fig=go.Figure(data=go.Violin(y=data_df['pred'], x=data_df['class'], line_color='black', 
+    full_fig=go.Figure(data=go.Violin(y=data_df['pred'], x=data_df['class'], line_color='#0D2A63', 
                                      meanline_visible=True, points=False, fillcolor=None, opacity=0.3, box=None,
                                      scalemode='count', showlegend = False))
 
@@ -86,24 +88,26 @@ def predicted_proba_violin_plot(true_y, predicted_proba, threshold_step = 0.01,
         titles[threshold] = titles[threshold][:-3] #removes last 3 char (2 spaces and comma)
 
         # NOTE: px strip generates n plots, one for each color class (TN, FP, FN, TP) it finds)
-        strip_points_fig = px.strip(data_df, x='class', y='pred', color= threshold_string, 
-                                    color_discrete_map = {'FN':'red', 'FP':'mediumpurple',
-                                                          'TP':'green', 'TN':'blue'},
+        strip_points_fig = px.strip(data_df, x='class', y='pred', color=threshold_string, 
+                                    color_discrete_map = {'FN':'#EF71D9', 'FP':'#EF553B',
+                                                          'TP':'#00CC96', 'TN':'#636EFA'},
                                     log_y=True, width=550, height=550, hover_data = [data_df.index])
 
-        strip_points_fig.update_traces(hovertemplate = 'Idx = %{customdata}<br>Class = %{x}<br>Pred = %{y}', jitter = 1, marker_size=3)
+        strip_points_fig.update_traces(hovertemplate = 'Idx = %{customdata}<br>Class = %{x}<br>Pred = %{y}', jitter = 1, marker_size=marker_size)
 
         length_fig_list.append(len(strip_points_fig.data))
 
         for i in range(len(strip_points_fig.data)):
             strip_points_fig.data[i].visible=False
-            full_fig.add_trace(strip_points_fig.data[i])
+
+        full_fig.add_traces(list(strip_points_fig.select_traces()))
 
     full_fig.update_layout(legend_font_size=9.5, legend_itemsizing='constant', legend_traceorder='grouped', 
                            title=dict(text = main_title + '<span style="font-size: 13px;">' \
                                       + titles[threshold_values[0]] + '</span>',
                                       y = 0.965, yanchor = 'bottom'), 
                            width=550, height=550)
+
     full_fig.update_layout(margin=dict(l=40, r=40, t=60, b=40))
 
     # makes visible the first strip points figure
@@ -201,11 +205,10 @@ def curve_PR_plot(true_y, predicted_proba, beta = 1, title = "Precision Recall C
                              "Recall":recall.tolist(),
                              "Precision":precision.tolist()})
 
-    pr_fig=px.line(curve_df, x="Recall", y="Precision", hover_data=["Thresholds"], title=main_title)
+    full_fig=px.line(curve_df, x="Recall", y="Precision", hover_data=["Thresholds"], title=main_title)
 
-    pr_fig.update_traces(hovertemplate='Threshold: %{customdata:.4f} <br>Precision: %{y:.4f} <br>Recall: %{x:.4f}<extra></extra>')
-    pr_fig.update_traces(line_color='#222A2A', line_width=2, textposition="top center")
-    full_fig = pr_fig
+    full_fig.update_traces(hovertemplate='Threshold: %{customdata:.4f} <br>Precision: %{y:.4f} <br>Recall: %{x:.4f}<extra></extra>')
+    full_fig.update_traces(textposition="top center") 
 
     f_scores = np.linspace(0.2, 0.8, num=4)
 
@@ -222,18 +225,19 @@ def curve_PR_plot(true_y, predicted_proba, beta = 1, title = "Precision Recall C
 
         iso_fig=px.line(recall_precision_df, x="recall", y="precision")
         iso_fig.update_traces(hovertemplate=[]) # no hover info displayed but keeps dashed lines
-        iso_fig.update_traces(line_color='#778AAE', line=dict(dash='dot'), line_width=0.3)
+        iso_fig.update_traces(line_color='#4C78A8', line=dict(dash='dot'), line_width=0.8)
 
         full_fig.add_annotation(x=0.90, y=y[45] + 0.01, text="f"+ str(beta) + "={0:0.1f}".format(f_score),
                                 showarrow=False,yshift=10)       
-        full_fig=go.Figure(data = full_fig.data + iso_fig.data, layout = full_fig.layout)
+
+        full_fig.add_traces(list(iso_fig.select_traces()))
 
     area_under_pr_curve = auc(recall, precision)
 
     full_fig.update_xaxes(range=[0.0, 1.0],title_text='Recall')
     full_fig.update_yaxes(range=[0.0, 1.05],title_text='Precision')
 
-    full_fig.add_shape(type='line', line=dict(dash='dash'),x0=0, x1=1, y0=baseline, y1=baseline)
+    full_fig.add_shape(type='line', line=dict(dash='dash', color = '#20313e'),x0=0, x1=1, y0=baseline, y1=baseline)
 
     full_fig['data'][0]['showlegend']= True
     full_fig['data'][1]['showlegend']= True
@@ -244,8 +248,8 @@ def curve_PR_plot(true_y, predicted_proba, beta = 1, title = "Precision Recall C
                            legend_font_size=9.5, 
                            width=550, height=550)
 
-    full_fig.update_xaxes(showspikes=True)
-    full_fig.update_yaxes(showspikes=True)
+    full_fig.update_xaxes(showspikes=True, spikedash = 'dot', spikethickness=2)
+    full_fig.update_yaxes(showspikes=True, spikedash = 'dot', spikethickness=2)
     full_fig.update_layout(margin=dict(l=40, r=40, t=40, b=40))
     full_fig.show()
 
@@ -292,22 +296,22 @@ def curve_ROC_plot(true_y, predicted_proba, title = "Receiver Operating Characte
                   hover_data=["Thresholds"],
                   width=550, height=550)
 
-    fig.update_traces(line_color="#222A2A", line_width=2, textposition="top center")
+    fig.update_traces(textposition="top center")
     fig.update_traces(hovertemplate='Threshold: %{customdata:.4f} <br>False Positive Rate: %{x:.4f} <br>True Positive Rate: %{y:.4f}<extra></extra>')
 
-    fig.add_shape(type="line", line=dict(dash="dash"), 
+    fig.add_shape(type="line", line=dict(dash="dash", color = '#20313e'), 
                   x0=0, x1=1, y0=0, y1=1)
 
     area_under_ROC_curve = auc(fpr, tpr)
 
     fig["data"][0]["name"]= f"ROC Curve (AUC={area_under_ROC_curve:.3f})"
-
     fig["data"][0]["showlegend"]= True
+
     fig.update_layout(legend = dict(yanchor="top", y=0.20, xanchor="left", x=0.5), 
                       legend_font_size=9.5)
 
-    fig.update_xaxes(showspikes=True)
-    fig.update_yaxes(showspikes=True)
+    fig.update_xaxes(showspikes=True, spikedash = 'dot', spikethickness=2)
+    fig.update_yaxes(showspikes=True, spikedash = 'dot', spikethickness=2)  
 
     fig.update_yaxes(scaleanchor="x", scaleratio=1)
     fig.update_xaxes(range=[0,1], constrain="domain")