Update DiagnosticReport to calculate base correctness of synthetic …

…data (#496)

sdmetrics/reports/base_report.py

@@ -25,7 +25,6 @@ def __init__(self):
         self._overall_score = None
         self.is_generated = False
         self._properties = {}
-        self._results_handler = None
         self.report_info = {
             'report_type': self.__class__.__name__,
             'generated_date': None,
@@ -91,6 +90,25 @@ def convert_datetimes(real_data, synthetic_data, metadata):
                 except Exception:
                     continue
 
+    def _print_results(self, verbose):
+        """Print the results.
+
+        Args:
+            verbose (bool):
+                Whether or not to print results to std.out.
+        """
+        if verbose:
+            sys.stdout.write(
+                f'\nOverall Score: {round(self._overall_score * 100, 2)}%\n\n'
+            )
+            sys.stdout.write('Properties:\n')
+
+            for property_name, property_instance in self._properties.items():
+                property_score = round(property_instance._compute_average() * 100, 2)
+                sys.stdout.write(
+                    f'- {property_name}: {property_score}%\n'
+                )
+
     def generate(self, real_data, synthetic_data, metadata, verbose=True):
         """Generate report.
 
@@ -152,7 +170,7 @@ def generate(self, real_data, synthetic_data, metadata, verbose=True):
         end_time = time.time()
         self.report_info['generation_time'] = end_time - start_time
 
-        self._handle_results(verbose)
+        self._print_results(verbose)
 
     def _check_property_name(self, property_name):
         """Check that the given property name is valid.
@@ -168,6 +186,16 @@ def _check_property_name(self, property_name):
                 f" Valid property names are '{valid_property_names}'."
             )
 
+    def get_score(self):
+        """Return the overall score.
+
+        Returns:
+            float
+                The overall score.
+        """
+        self._check_report_generated()
+        return self._overall_score
+
     def get_info(self):
         """Get the information about the report."""
         return deepcopy(self.report_info)

sdmetrics/reports/multi_table/diagnostic_report.py

@@ -1,36 +1,19 @@
 """Multi table diagnostic report."""
-from copy import deepcopy
-
-from sdmetrics.reports._results_handler import DiagnosticReportResultsHandler
-from sdmetrics.reports.multi_table._properties import Boundary, Coverage, Synthesis
+from sdmetrics.reports.multi_table._properties import DataValidity, RelationshipValidity, Structure
 from sdmetrics.reports.multi_table.base_multi_table_report import BaseMultiTableReport
 
 
 class DiagnosticReport(BaseMultiTableReport):
     """Multi table diagnostic report.
 
     This class creates a diagnostic report for multi-table data. It calculates the diagnostic
-    score along three properties - Synthesis, Coverage, and Boundary.
+    score along three properties - Relationship Validity, Data Structure, and Data Validity.
     """
 
     def __init__(self):
         super().__init__()
         self._properties = {
-            'Coverage': Coverage(),
-            'Boundary': Boundary(),
-            'Synthesis': Synthesis()
+            'Data Validity': DataValidity(),
+            'Data Structure': Structure(),
+            'Relationship Validity': RelationshipValidity()
         }
-        self._results_handler = DiagnosticReportResultsHandler()
-
-    def _handle_results(self, verbose):
-        self._results_handler.print_results(self._properties, verbose)
-
-    def get_results(self):
-        """Return the diagnostic results.
-
-        Returns:
-            dict
-                The diagnostic results.
-        """
-        self._check_report_generated()
-        return deepcopy(self._results_handler.results)

sdmetrics/reports/multi_table/quality_report.py

@@ -1,5 +1,4 @@
 """Multi table quality report."""
-from sdmetrics.reports._results_handler import QualityReportResultsHandler
 from sdmetrics.reports.multi_table._properties import (
     Cardinality, ColumnPairTrends, ColumnShapes, InterTableTrends)
 from sdmetrics.reports.multi_table.base_multi_table_report import BaseMultiTableReport
@@ -20,17 +19,3 @@ def __init__(self):
             'Cardinality': Cardinality(),
             'Intertable Trends': InterTableTrends()
         }
-        self._results_handler = QualityReportResultsHandler()
-
-    def _handle_results(self, verbose):
-        self._results_handler.print_results(self._properties, self._overall_score, verbose)
-
-    def get_score(self):
-        """Return the overall quality score.
-
-        Returns:
-            float
-                The overall quality score.
-        """
-        self._check_report_generated()
-        return self._overall_score

sdmetrics/reports/single_table/diagnostic_report.py

@@ -1,39 +1,18 @@
 """Single table diagnostic report."""
-import logging
-from copy import deepcopy
-
-from sdmetrics.reports._results_handler import DiagnosticReportResultsHandler
 from sdmetrics.reports.base_report import BaseReport
-from sdmetrics.reports.single_table._properties import Boundary, Coverage, Synthesis
-
-LOGGER = logging.getLogger(__name__)
+from sdmetrics.reports.single_table._properties import DataValidity, Structure
 
 
 class DiagnosticReport(BaseReport):
     """Single table diagnostic report.
 
     This class creates a diagnostic report for single-table data. It calculates the diagnostic
-    score along three properties - Synthesis, Coverage, and Boundary.
+    score along two properties - Data Structure and Data Validity.
     """
 
     def __init__(self):
         super().__init__()
         self._properties = {
-            'Coverage': Coverage(),
-            'Boundary': Boundary(),
-            'Synthesis': Synthesis()
+            'Data Validity': DataValidity(),
+            'Data Structure': Structure(),
         }
-        self._results_handler = DiagnosticReportResultsHandler()
-
-    def _handle_results(self, verbose):
-        self._results_handler.print_results(self._properties, verbose)
-
-    def get_results(self):
-        """Return the diagnostic results.
-
-        Returns:
-            dict
-                The diagnostic results.
-        """
-        self._check_report_generated()
-        return deepcopy(self._results_handler.results)

sdmetrics/reports/single_table/quality_report.py

@@ -1,5 +1,4 @@
 """Single table quality report."""
-from sdmetrics.reports._results_handler import QualityReportResultsHandler
 from sdmetrics.reports.base_report import BaseReport
 from sdmetrics.reports.single_table._properties import ColumnPairTrends, ColumnShapes
 
@@ -17,17 +16,3 @@ def __init__(self):
             'Column Shapes': ColumnShapes(),
             'Column Pair Trends': ColumnPairTrends()
         }
-        self._results_handler = QualityReportResultsHandler()
-
-    def _handle_results(self, verbose):
-        self._results_handler.print_results(self._properties, self._overall_score, verbose)
-
-    def get_score(self):
-        """Return the overall quality score.
-
-        Returns:
-            float
-                The overall quality score.
-        """
-        self._check_report_generated()
-        return self._overall_score