Filled out preserve method, including error handling. Refactor on spec test.

Author: interpret-ml

src/python/interpret/test/test_develop.py

@@ -15,4 +15,3 @@ def test_print_debug_info():
     # Very light check, just testing if the function runs.
     print_debug_info()
     assert 1 == 1
-

src/python/interpret/test/test_explainers.py

@@ -1,47 +1,15 @@
 # Copyright (c) 2019 Microsoft Corporation
 # Distributed under the MIT software license
 
-from ..data import ClassHistogram
-from ..perf import ROC, RegressionPerf
 
-from ..blackbox import LimeTabular
-from ..blackbox import ShapKernel
-from ..blackbox import MorrisSensitivity
-from ..blackbox import PartialDependence
-
-from ..glassbox import LogisticRegression, LinearRegression
-from ..glassbox import ClassificationTree, RegressionTree
-from ..glassbox import DecisionListClassifier
-from ..glassbox import ExplainableBoostingClassifier, ExplainableBoostingRegressor
-
-from .utils import synthetic_classification
+from .utils import synthetic_classification, get_all_explainers
 from .utils import assert_valid_explanation, assert_valid_model_explainer
 
+from ..glassbox import LogisticRegression
 
-def test_spec_synthetic():
-    data_explainer_classes = [ClassHistogram]
-    perf_explainer_classes = [ROC, RegressionPerf]
-    model_explainer_classes = [
-        ClassificationTree,
-        DecisionListClassifier,
-        LogisticRegression,
-        ExplainableBoostingClassifier,
-        RegressionTree,
-        LinearRegression,
-        ExplainableBoostingRegressor,
-    ]
-    blackbox_explainer_classes = [
-        LimeTabular,
-        ShapKernel,
-        MorrisSensitivity,
-        PartialDependence,
-    ]
-    all_explainers = []
-    all_explainers.extend(model_explainer_classes)
-    all_explainers.extend(blackbox_explainer_classes)
-    all_explainers.extend(data_explainer_classes)
-    all_explainers.extend(perf_explainer_classes)
 
+def test_spec_synthetic():
+    all_explainers = get_all_explainers()
     data = synthetic_classification()
     blackbox = LogisticRegression()
     blackbox.fit(data["train"]["X"], data["train"]["y"])

src/python/interpret/test/test_interactive.py

@@ -1,5 +1,6 @@
 # Copyright (c) 2019 Microsoft Corporation
 # Distributed under the MIT software license
+# TODO: Testing for show/snap functions.
 
 from ..visual.interactive import set_show_addr, get_show_addr, shutdown_show_server
 

src/python/interpret/test/utils.py

@@ -1,6 +1,19 @@
 # Copyright (c) 2019 Microsoft Corporation
 # Distributed under the MIT software license
 
+from ..data import ClassHistogram
+from ..perf import ROC, RegressionPerf
+
+from ..blackbox import LimeTabular
+from ..blackbox import ShapKernel
+from ..blackbox import MorrisSensitivity
+from ..blackbox import PartialDependence
+
+from ..glassbox import LogisticRegression, LinearRegression
+from ..glassbox import ClassificationTree, RegressionTree
+from ..glassbox import DecisionListClassifier
+from ..glassbox import ExplainableBoostingClassifier, ExplainableBoostingRegressor
+
 import pandas as pd
 import numpy as np
 from sklearn.model_selection import train_test_split
@@ -10,6 +23,33 @@
 from sklearn.base import is_classifier
 
 
+def get_all_explainers():
+    data_explainer_classes = [ClassHistogram]
+    perf_explainer_classes = [ROC, RegressionPerf]
+    model_explainer_classes = [
+        ClassificationTree,
+        DecisionListClassifier,
+        LogisticRegression,
+        ExplainableBoostingClassifier,
+        RegressionTree,
+        LinearRegression,
+        ExplainableBoostingRegressor,
+    ]
+    blackbox_explainer_classes = [
+        LimeTabular,
+        ShapKernel,
+        MorrisSensitivity,
+        PartialDependence,
+    ]
+    all_explainers = []
+    all_explainers.extend(model_explainer_classes)
+    all_explainers.extend(blackbox_explainer_classes)
+    all_explainers.extend(data_explainer_classes)
+    all_explainers.extend(perf_explainer_classes)
+
+    return all_explainers
+
+
 def synthetic_regression():
     dataset = _synthetic("regression")
     return dataset

src/python/interpret/visual/interactive.py

@@ -133,7 +133,7 @@ def preserve(explanation, selector_key=None, file_name=None, **kwargs):
 
     If file_name is not None the following occurs:
     - For Plotly figures, saves to HTML using `plot`.
-    - For dataframes, saves to CSV using `to_csv`.
+    - For dataframes, saves to HTML using `to_html`.
     - For strings (html), saves to HTML.
     - For Dash components, fails with exception. This is currently not supported.
 
@@ -147,17 +147,58 @@ def preserve(explanation, selector_key=None, file_name=None, **kwargs):
         None.
     """
 
+    try:
+        # Get explanation key
+        if selector_key is None:
+            key = None
+        else:
+            series = explanation.selector[explanation.selector.columns[0]]
+            key = series[series == selector_key].index[0]
+
+        # Get visual object
+        visual = explanation.visualize(key=key)
+
+        # Output to front-end/file
+        _preserve_output(
+            explanation.name,
+            visual,
+            selector_key=selector_key,
+            file_name=file_name,
+            **kwargs
+        )
+        return None
+    except Exception as e:
+        log.error(e, exc_info=True)
+        raise e
+
+
+def _preserve_output(
+    explanation_name, visual, selector_key=None, file_name=None, **kwargs
+):
     from plotly.offline import iplot, plot, init_notebook_mode
-    from IPython.display import display, HTML
+    from IPython.display import display, display_html
+    from base64 import b64encode
+
     init_notebook_mode(connected=True)
 
-    if selector_key is None:
-        key = None
-    else:
-        series = explanation.selector[explanation.selector.columns[0]]
-        key = series[series == selector_key].index[0]
+    def render_html(html_string):
+        base64_html = b64encode(html_string.encode("utf-8")).decode("ascii")
+        final_html = """<iframe src="data:text/html;base64,{data}" width="100%" height=400 frameBorder="0"></iframe>""".format(
+            data=base64_html
+        )
+        display_html(final_html, raw=True)
+
+    if visual is None:
+        msg = "No visualization for explanation [{0}] with selector_key [{1}]".format(
+            explanation_name, selector_key
+        )
+        log.error(msg)
+        if file_name is None:
+            render_html(msg)
+        else:
+            pass
+        return False
 
-    visual = explanation.visualize(key=key)
     if isinstance(visual, go.Figure):
         if file_name is None:
             iplot(visual, **kwargs)
@@ -167,18 +208,24 @@ def preserve(explanation, selector_key=None, file_name=None, **kwargs):
         if file_name is None:
             display(visual, **kwargs)
         else:
-            visual.to_csv(file_name, **kwargs)
+            visual.to_html(file_name, **kwargs)
     elif isinstance(visual, str):
         if file_name is None:
-            with(file_name, "w") as f:
-                f.write(visual)
+            render_html(visual)
         else:
-            HTML(visual, **kwargs)
+            with open(file_name, "w") as f:
+                f.write(visual)
     elif isinstance(visual, dash_base.Component):
         msg = "Preserving dash components is currently not supported."
-        raise Exception(msg)
+        if file_name is None:
+            render_html(msg)
+        log.error(msg)
+        return False
     else:
         msg = "Visualization cannot be preserved for type: {0}.".format(type(visual))
-        raise Exception(msg)
+        if file_name is None:
+            render_html(msg)
+        log.error(msg)
+        return False
 
-    return None
+    return True