[MLC-413] replaced mathplot with plotly (#312)

* replaced mathplot with plotly

deepchecks/checks/__init__.py

@@ -51,7 +51,6 @@
 )
 
 from .performance import (
-
     PerformanceReport,
     ConfusionMatrixReport,
     RocReport,

deepchecks/checks/performance/regression_error_distribution.py

@@ -9,7 +9,7 @@
 # ----------------------------------------------------------------------------
 #
 """The regression_error_distribution check module."""
-import matplotlib.pyplot as plt
+import plotly.express as px
 import pandas as pd
 from scipy.stats import kurtosis
 from sklearn.base import BaseEstimator
@@ -69,15 +69,18 @@ def _regression_error_distribution(self, dataset: Dataset, model: BaseEstimator)
         n_smallest_diff.name = n_smallest_diff.name + ' Prediction Difference'
         n_smallest = pd.concat([dataset.data.loc[n_smallest_diff.index], n_smallest_diff], axis=1)
 
-        def display_hist():
-            diff = y_test - y_pred
-            diff.hist(bins=40)
-            plt.title('Histogram of prediction errors')
-            plt.xlabel(f'{dataset.label_name} prediction error')
-            plt.ylabel('Count')
+        display = [
+            px.histogram(
+                x=diff.values,
+                nbins=40,
+                title='Histogram of prediction errors',
+                labels={'x': f'{dataset.label_name} prediction error', 'y': 'Count'},
+                width=700, height=500
+            ),
+            'Largest over estimation errors:', n_largest,
+            'Largest under estimation errors:', n_smallest
+        ]
 
-        display = [display_hist, 'Largest over estimation errors:', n_largest,
-                   'Largest under estimation errors:', n_smallest]
         return CheckResult(value=kurtosis_value, display=display)
 
     def add_condition_kurtosis_not_less_than(self, min_kurtosis: float = -0.1):

deepchecks/checks/performance/regression_systematic_error.py

@@ -9,7 +9,7 @@
 # ----------------------------------------------------------------------------
 #
 """The RegressionSystematicError check module."""
-import matplotlib.pyplot as plt
+import plotly.graph_objects as go
 from sklearn.base import BaseEstimator
 from sklearn.metrics import mean_squared_error
 
@@ -51,16 +51,25 @@ def _regression_error_distribution(self, dataset: Dataset, model: BaseEstimator)
         diff = y_test - y_pred
         diff_mean = diff.mean()
 
-        def display_box_plot():
-            red_square = dict(markerfacecolor='r', marker='s')
-            _, ax = plt.subplots()
-            ax.set_title('Box plot of the model prediction error')
-            ax.boxplot(diff, vert=False, flierprops=red_square)
-            ax.axvline(x=diff_mean, linestyle='--')
-            ax.annotate(xy=(diff_mean + 0.01, 1.2), text='mean error')
-
-        display = ['Non-zero mean of the error distribution indicated the presents of \
-            systematic error in model predictions', display_box_plot]
+        display = [
+            'Non-zero mean of the error distribution indicated the presents '
+            'of systematic error in model predictions',
+            go.Figure()
+            .add_trace(go.Box(
+                x=diff.values,
+                name='Model Prediction Error',
+                boxpoints='suspectedoutliers',
+                marker=dict(outliercolor='red'),
+                hoverinfo='x',
+                hoveron='points'))
+            .update_layout(
+                title_text='Box plot of the model prediction error',
+                width=800, height=500)
+            .add_vline(
+                x=diff_mean + 0.01,
+                line_dash='dash',
+                annotation_text='Mean error')
+        ]
 
         return CheckResult(value={'rmse': rmse, 'mean_error': diff_mean}, display=display)