Add dataset import differ utility

run_tests.sh

@@ -17,7 +17,7 @@
 set -e
 
 # Array of top-level folders with Python code.
-PYTHON_FOLDERS="util/ scripts/ import-automation/executor"
+PYTHON_FOLDERS="util/ scripts/ import-automation/executor tools/"
 
 # Flag used signal if Python requirements have already been installed.
 PYTHON_REQUIREMENTS_INSTALLED=false

tools/differ/README.md

@@ -0,0 +1,29 @@
+# Dataset Differ
+
+This utility generates a diff (point and series analysis) of two versions of the same dataset for import analysis.
+
+**Usage**
+```
+python differ.py --current_data=<path> --previous_data=<path>
+```
+
+Parameter description
+current_data: Path to the current MCF data (single mcf file or folder/* supported).
+previous_data: Path to the previous MCF data (single mcf file or folder/* supported).
+output_location: Path to the output data folder.
+groupby_columns: Columns to group data for diff analysis in the order var,place,time etc. Default value: “variableMeasured,observationAbout,observationDate”
+value_columns: Columns with statvar value (unit etc.) for diff analysis. Default value:  "value,unit"
+
+Summary output generated is of the form below showing counts of differences for each variable.
+
+variableMeasured   added  deleted  modified  same  total
+0   dcid:var1       1      0       0          0     1
+1   dcid:var2       0      2       1          1     4
+2   dcid:var3       0      0       1          0     1
+3   dcid:var4       0      2       0          0     2
+
+Detailed diff output is written to files for further analysis.
+- point-analysis-summary.csv: diff summry for point analysis
+- point-analysis-results.csv: detailed results for point analysis
+- series-analysis-summary.csv: diff summry for series analysis
+- series-analysis-results.csv: detailed results for series analysis

tools/differ/differ.py

@@ -0,0 +1,251 @@
+# Copyright 2020 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Utility to generate a dataset diff for import analysis."""
+
+import os
+import pandas as pd
+import random
+
+from absl import app
+from absl import flags
+from absl import logging
+
+import helper
+
+FLAGS = flags.FLAGS
+flags.DEFINE_string(
+    'current_data', '', 'Path to the current MCF data \
+  (single mcf file or folder/* supported).')
+flags.DEFINE_string(
+    'previous_data', '', 'Path to the previous MCF data \
+  (single mcf file or folder/* supported).')
+flags.DEFINE_string('output_location', 'results', \
+  'Path to the output data folder.')
+
+flags.DEFINE_string(
+    'groupby_columns', 'variableMeasured,observationAbout,observationDate',
+    'Columns to group data for diff analysis in the order (var,place,time etc.).'
+)
+flags.DEFINE_string(
+    'value_columns', 'value,unit',
+    'Columns with statvar value (unit etc.) for diff analysis.')
+
+SAMPLE_COUNT = 3
+
+
+class DatasetDiffer:
+    """
+  Utility to generate a diff (point and series analysis) 
+  of two versions of the same dataset for import analysis. 
+
+  Usage:
+  $ python differ.py --current_data=<path> --previous_data=<path>
+
+  Summary output generated is of the form below showing 
+  counts of differences for each variable.  
+
+  variableMeasured   added  deleted  modified  same  total
+  0   dcid:var1       1      0       0          0     1
+  1   dcid:var2       0      2       1          1     4
+  2   dcid:var3       0      0       1          0     1
+  3   dcid:var4       0      2       0          0     2
+
+  Detailed diff output is written to files for further analysis.
+  - point-analysis-summary.csv: diff summry for point analysis
+  - point-analysis-results.csv: detailed results for point analysis
+  - series-analysis-summary.csv: diff summry for series analysis
+  - series-analysis-results.csv: detailed results for series analysis
+
+  """
+
+    def __init__(self, groupby_columns, value_columns):
+        self.groupby_columns = groupby_columns.split(',')
+        self.value_columns = value_columns.split(',')
+        self.variable_column = self.groupby_columns[0]
+        self.place_column = self.groupby_columns[1]
+        self.time_column = self.groupby_columns[2]
+        self.diff_column = '_diff_result'
+
+    def _cleanup_data(self, df: pd.DataFrame):
+        for column in ['added', 'deleted', 'modified', 'same']:
+            df[column] = df[column] if column in df.columns else 0
+            df[column] = df[column].fillna(0).astype(int)
+
+    def _get_samples(self, row):
+        years = sorted(row[self.time_column])
+        if len(years) > SAMPLE_COUNT:
+            return years[0] + random.sample(years[1:-1],
+                                            SAMPLE_COUNT - 2) + years[-1]
+        else:
+            return years
+
+    # Processes two dataset files to identify changes.
+    def process_data(self, previous_df: pd.DataFrame,
+                     current_df: pd.DataFrame) -> pd.DataFrame:
+        """ 
+        Process previous and current datasets to generate 
+        the intermediate data for point and series analysis.
+        Args:
+          current_df: dataframe with current (new) data
+          previous_df: dataframe with previous (old) data
+        Returns:
+          intermediate merged data for analysis
+        """
+        cur_df_columns = current_df.columns.values.tolist()
+        self.groupby_columns = [
+            i for i in self.groupby_columns if i in cur_df_columns
+        ]
+        self.value_columns = [
+            i for i in self.value_columns if i in cur_df_columns
+        ]
+        df1 = previous_df.loc[:, self.groupby_columns + self.value_columns]
+        df2 = current_df.loc[:, self.groupby_columns + self.value_columns]
+        df1['_value_combined'] = df1[self.value_columns]\
+          .apply(lambda row: '_'.join(row.values.astype(str)), axis=1)
+        df2['_value_combined'] = df2[self.value_columns]\
+          .apply(lambda row: '_'.join(row.values.astype(str)), axis=1)
+        df1.drop(columns=self.value_columns, inplace=True)
+        df2.drop(columns=self.value_columns, inplace=True)
+        # Perform outer join operation to identify differences.
+        result = pd.merge(df1,
+                          df2,
+                          on=self.groupby_columns,
+                          how='outer',
+                          indicator=self.diff_column)
+        result[self.diff_column] = result.apply(
+          lambda row: 'added' if row[self.diff_column] == 'right_only' \
+          else 'deleted' if row[self.diff_column] == 'left_only' \
+          else 'modified' if row['_value_combined_x'] != row['_value_combined_y'] \
+          else 'same', axis=1)
+        return result
+
+    def point_analysis(self,
+                       in_data: pd.DataFrame) -> (pd.DataFrame, pd.DataFrame):
+        """ 
+        Performs point diff analysis to identify data point changes.
+        Args:
+          in_data: intermediate data generated by processing previous/current data
+        Returns:
+          summary and results from the analysis
+        """
+        column_list = [
+            self.variable_column, self.place_column, self.time_column,
+            self.diff_column
+        ]
+        result = in_data.loc[:, column_list]
+        result = result.groupby(
+            [self.variable_column, self.diff_column],
+            observed=True,
+            as_index=False)[[self.place_column,
+                             self.time_column]].agg(lambda x: x.tolist())
+        result['size'] = result.apply(lambda row: len(row[self.place_column]),
+                                      axis=1)
+        result[self.place_column] = result.apply(lambda row: random.sample(
+            row[self.place_column],
+            min(SAMPLE_COUNT, len(row[self.place_column]))),
+                                                 axis=1)
+        result[self.time_column] = result.apply(self._get_samples, axis=1)
+        summary = result.pivot(
+          index=self.variable_column, columns=self.diff_column, values='size')\
+          .reset_index().rename_axis(None, axis=1)
+        self._cleanup_data(summary)
+        summary['total'] = summary.apply(lambda row: row['added'] + row[
+            'deleted'] + row['modified'] + row['same'],
+                                         axis=1)
+        return summary, result
+
+    def series_analysis(self,
+                        in_data: pd.DataFrame) -> (pd.DataFrame, pd.DataFrame):
+        """ 
+        Performs series diff analysis to identify time series changes.
+        Args:
+          in_data: intermediate data generated by processing previous/current data
+        Returns:
+          summary and results from the analysis
+        """
+        column_list = [
+            self.variable_column, self.place_column, self.diff_column
+        ]
+        result = in_data.loc[:, column_list]
+        result = result.groupby(column_list, as_index=False).size()
+        result = result.pivot(
+          index=[self.variable_column, self.place_column], columns=self.diff_column, values='size')\
+          .reset_index().rename_axis(None, axis=1)
+        self._cleanup_data(result)
+        result[self.diff_column] = result.apply(lambda row: 'added' if row['added'] > 0 \
+          and row['deleted'] + row['modified'] + row['same'] == 0 \
+          else 'deleted' if row['deleted'] > 0 and row['added'] + row['modified'] + row['same'] == 0 \
+          else 'modified' if row['deleted'] > 0 or row['added'] > 0 or row['modified'] > 0 \
+          else 'same', axis=1)
+        result = result[column_list]
+        result = result.groupby(
+            [self.variable_column, self.diff_column],
+            observed=True,
+            as_index=False)[self.place_column].agg(lambda x: x.tolist())
+        result['size'] = result.apply(lambda row: len(row[self.place_column]),
+                                      axis=1)
+        result[self.place_column] = result.apply(lambda row: random.sample(
+            row[self.place_column],
+            min(SAMPLE_COUNT, len(row[self.place_column]))),
+                                                 axis=1)
+        summary = result.pivot(
+          index=self.variable_column, columns=self.diff_column, values='size')\
+          .reset_index().rename_axis(None, axis=1)
+        self._cleanup_data(summary)
+        summary['total'] = summary.apply(lambda row: row['added'] + row[
+            'deleted'] + row['modified'] + row['same'],
+                                         axis=1)
+        return summary, result
+
+
+def main(_):
+    '''Runs the differ.'''
+    differ = DatasetDiffer(FLAGS.groupby_columns, FLAGS.value_columns)
+
+    if not os.path.exists(FLAGS.output_location):
+        os.makedirs(FLAGS.output_location)
+    logging.info('Loading data...')
+    current_df = helper.load_data(FLAGS.current_data, FLAGS.output_location)
+    previous_df = helper.load_data(FLAGS.previous_data, FLAGS.output_location)
+
+    logging.info('Processing data...')
+    in_data = differ.process_data(previous_df, current_df)
+
+    logging.info('Point analysis:')
+    summary, result = differ.point_analysis(in_data)
+    result.sort_values(by=[differ.diff_column, differ.variable_column],
+                       inplace=True)
+    print(summary.head(10))
+    print(result.head(10))
+    helper.write_data(summary, FLAGS.output_location,
+                      'point-analysis-summary.csv')
+    helper.write_data(result, FLAGS.output_location,
+                      'point-analysis-results.csv')
+
+    logging.info('Series analysis:')
+    summary, result = differ.series_analysis(in_data)
+    result.sort_values(by=[differ.diff_column, differ.variable_column],
+                       inplace=True)
+    print(summary.head(10))
+    print(result.head(10))
+    helper.write_data(summary, FLAGS.output_location,
+                      'series-analysis-summary.csv')
+    helper.write_data(result, FLAGS.output_location,
+                      'series-analysis-results.csv')
+
+    logging.info('Differ output written to folder: %s', FLAGS.output_location)
+
+
+if __name__ == '__main__':
+    app.run(main)

tools/differ/differ_test.py

@@ -0,0 +1,54 @@
+# Copyright 2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import pandas as pd
+import unittest
+
+from pandas.testing import assert_frame_equal
+from differ import DatasetDiffer
+
+import helper
+
+module_dir = os.path.dirname(__file__)
+
+
+class TestDiffer(unittest.TestCase):
+    '''
+  Test Class to compare expected output in test/ directory to the
+  output generated by DatasetDiffer class
+  '''
+
+    def test_diff_analysis(self):
+        groupby_columns = 'variableMeasured,observationAbout,observationDate'
+        value_columns = 'value'
+
+        differ = DatasetDiffer(groupby_columns, value_columns)
+        current = helper.load_mcf_file(
+            os.path.join(module_dir, 'test', 'current.mcf'))
+        previous = helper.load_mcf_file(
+            os.path.join(module_dir, 'test', 'previous.mcf'))
+
+        in_data = differ.process_data(previous, current)
+        summary, result = differ.point_analysis(in_data)
+        result = pd.read_csv(os.path.join(module_dir, 'test', 'result1.csv'))
+        assert_frame_equal(summary, result)
+
+        summary, result = differ.series_analysis(in_data)
+        result = pd.read_csv(os.path.join(module_dir, 'test', 'result2.csv'))
+        assert_frame_equal(summary, result)
+
+
+if __name__ == '__main__':
+    unittest.main()