Support more than one statistic for summarizing time lists

issue_metrics.py

@@ -32,14 +32,14 @@
 from classes import IssueWithMetrics
 from discussions import get_discussions
 from json_writer import write_to_json
-from labels import get_average_time_in_labels, get_label_metrics
+from labels import get_stats_time_in_labels, get_label_metrics
 from markdown_writer import write_to_markdown
-from time_to_answer import get_average_time_to_answer, measure_time_to_answer
-from time_to_close import get_average_time_to_close, measure_time_to_close
+from time_to_answer import get_stats_time_to_answer, measure_time_to_answer
+from time_to_close import get_stats_time_to_close, measure_time_to_close
 from time_to_ready_for_review import get_time_to_ready_for_review
 from time_to_merge import measure_time_to_merge
 from time_to_first_response import (
-    get_average_time_to_first_response,
+    get_stats_time_to_first_response,
     measure_time_to_first_response,
 )
 
@@ -317,36 +317,36 @@ def main():
         ignore_users=ignore_users,
     )
 
-    average_time_to_first_response = get_average_time_to_first_response(
+    stats_time_to_first_response = get_stats_time_to_first_response(
         issues_with_metrics
     )
-    average_time_to_close = None
+    stats_time_to_close = None
     if num_issues_closed > 0:
-        average_time_to_close = get_average_time_to_close(issues_with_metrics)
+        stats_time_to_close = get_stats_time_to_close(issues_with_metrics)
 
-    average_time_to_answer = get_average_time_to_answer(issues_with_metrics)
+    stats_time_to_answer = get_stats_time_to_answer(issues_with_metrics)
 
-    # Get the average time in label for each label and store it in a dictionary
+    # Get stats describing the time in label for each label and store it in a dictionary
     # where the key is the label and the value is the average time
-    average_time_in_labels = get_average_time_in_labels(issues_with_metrics, labels)
+    stats_time_in_labels = get_stats_time_in_labels(issues_with_metrics, labels)
 
     # Write the results to json and a markdown file
     write_to_json(
         issues_with_metrics,
-        average_time_to_first_response,
-        average_time_to_close,
-        average_time_to_answer,
-        average_time_in_labels,
+        stats_time_to_first_response,
+        stats_time_to_close,
+        stats_time_to_answer,
+        stats_time_in_labels,
         num_issues_open,
         num_issues_closed,
         search_query,
     )
     write_to_markdown(
         issues_with_metrics,
-        average_time_to_first_response,
-        average_time_to_close,
-        average_time_to_answer,
-        average_time_in_labels,
+        stats_time_to_first_response,
+        stats_time_to_close,
+        stats_time_to_answer,
+        stats_time_in_labels,
         num_issues_open,
         num_issues_closed,
         labels,

labels.py

@@ -3,6 +3,7 @@
 from typing import List
 
 import github3
+import numpy
 import pytz
 
 from classes import IssueWithMetrics
@@ -88,32 +89,39 @@ def get_label_metrics(issue: github3.issues.Issue, labels: List[str]) -> dict:
     return label_metrics
 
 
-def get_average_time_in_labels(
+def get_stats_time_in_labels(
     issues_with_metrics: List[IssueWithMetrics],
     labels: List[str],
 ) -> dict[str, timedelta]:
-    """Calculate the average time spent in each label."""
-    average_time_in_labels = {}
-    number_of_issues_in_labels = {}
+    """Calculate stats describing time spent in each label."""
+    time_in_labels = {}
     for issue in issues_with_metrics:
         if issue.label_metrics:
             for label in issue.label_metrics:
                 if issue.label_metrics[label] is None:
                     continue
-                if label not in average_time_in_labels:
-                    average_time_in_labels[label] = issue.label_metrics[label]
-                    number_of_issues_in_labels[label] = 1
+                if label not in time_in_labels:
+                    time_in_labels[label] = [issue.label_metrics[label]]
                 else:
-                    average_time_in_labels[label] += issue.label_metrics[label]
-                    number_of_issues_in_labels[label] += 1
+                    time_in_labels[label].append(issue.label_metrics[label])
 
-    for label in average_time_in_labels:
-        average_time_in_labels[label] = (
-            average_time_in_labels[label] / number_of_issues_in_labels[label]
-        )
+    average_time_in_labels = {}
+    med_time_in_labels = {}
+    ninety_percentile_in_labels = {}
+    for label, time_list in time_in_labels.items():
+        average_time_in_labels[label] = numpy.average(time_list)
+        med_time_in_labels[label] = numpy.median(time_list)
+        ninety_percentile_in_labels[label] = numpy.percentile(time_list, 90, axis=0)
 
     for label in labels:
         if label not in average_time_in_labels:
             average_time_in_labels[label] = None
-
-    return average_time_in_labels
+            med_time_in_labels[label] = None
+            ninety_percentile_in_labels[label] = None
+
+    stats = {
+                 'avg': average_time_in_labels,
+                 'med': med_time_in_labels,
+                 '90p': ninety_percentile_in_labels
+    }
+    return stats

markdown_writer.py

@@ -170,32 +170,57 @@ def write_to_markdown(
 
 def write_overall_metrics_table(
     issues_with_metrics,
-    average_time_to_first_response,
-    average_time_to_close,
-    average_time_to_answer,
-    average_time_in_labels,
+    stats_time_to_first_response,
+    stats_time_to_close,
+    stats_time_to_answer,
+    stats_time_in_labels,
     num_issues_opened,
     num_issues_closed,
     labels,
     columns,
     file,
 ):
     """Write the overall metrics table to the markdown file."""
-    file.write("| Metric | Value |\n")
-    file.write("| --- | ---: |\n")
+    file.write("| Metric | Average | Median | 90th percentile |\n")
+    file.write("| --- | --- | --- | ---: |\n")
     if "Time to first response" in columns:
-        file.write(
-            f"| Average time to first response | {average_time_to_first_response} |\n"
-        )
+        if stats_time_to_first_response is not None:
+            file.write(
+                f"| Time to first response "
+                f"| {stats_time_to_first_response['avg']} "
+                f"| {stats_time_to_first_response['med']} "
+                f"| {stats_time_to_first_response['90p']} |\n"
+            )
+        else:
+            file.write("| Time to first response | None | None | None |\n")
     if "Time to close" in columns:
-        file.write(f"| Average time to close | {average_time_to_close} |\n")
+        if stats_time_to_close is not None:
+            file.write(
+                f"| Time to close "
+                f"| {stats_time_to_close['avg']} "
+                f"| {stats_time_to_close['med']} "
+                f"| {stats_time_to_close['90p']} |\n"
+            )
+        else:
+            file.write("| Time to close | None | None | None |\n")
     if "Time to answer" in columns:
-        file.write(f"| Average time to answer | {average_time_to_answer} |\n")
-    if labels and average_time_in_labels:
+        if stats_time_to_answer is not None:
+            file.write(
+                f"| Time to answer "
+                f"| {stats_time_to_answer['avg']} "
+                f"| {stats_time_to_answer['med']} "
+                f"| {stats_time_to_answer['90p']} |\n"
+            )
+        else:
+            file.write("| Time to answer | None | None | None |\n")
+    if labels and stats_time_in_labels:
         for label in labels:
-            if f"Time spent in {label}" in columns and label in average_time_in_labels:
+            if f"Time spent in {label}" in columns and label in stats_time_in_labels['avg']:
                 file.write(
-                    f"| Average time spent in {label} | {average_time_in_labels[label]} |\n"
+                    f"| Time spent in {label} "
+                    f"| {stats_time_in_labels['avg'][label]} "
+                    f"| {stats_time_in_labels['med'][label]} "
+                    f"| {stats_time_in_labels['90p'][label]} |\n"
                 )
     file.write(f"| Number of items that remain open | {num_issues_opened} |\n")
     file.write(f"| Number of items closed | {num_issues_closed} |\n")

requirements.txt

@@ -6,3 +6,4 @@ pytest==7.4.2
 pytest-cov==4.1.0
 flake8==6.1.0
 pylint==3.0.1
+numpy==1.26.0

test_issue_metrics.py

@@ -133,7 +133,7 @@ class TestMain(unittest.TestCase):
     @patch("issue_metrics.auth_to_github")
     @patch("issue_metrics.search_issues")
     @patch("issue_metrics.measure_time_to_first_response")
-    @patch("issue_metrics.get_average_time_to_first_response")
+    @patch("issue_metrics.get_stats_time_to_first_response")
     @patch.dict(
         os.environ,
         {
@@ -142,7 +142,7 @@ class TestMain(unittest.TestCase):
     )
     def test_main(
         self,
-        mock_get_average_time_to_first_response,
+        mock_get_stats_time_to_first_response,
         mock_measure_time_to_first_response,
         mock_search_issues,
         mock_auth_to_github,
@@ -179,10 +179,10 @@ def test_main(
         ]
         mock_measure_time_to_first_response.return_value = mock_issues_with_ttfr
 
-        # Set up the mock get_average_time_to_first_response function
-        mock_average_time_to_first_response = 15
-        mock_get_average_time_to_first_response.return_value = (
-            mock_average_time_to_first_response
+        # Set up the mock get_stats_time_to_first_response function
+        mock_stats_time_to_first_response = 15
+        mock_get_stats_time_to_first_response.return_value = (
+            mock_stats_time_to_first_response
         )
 
         # Call main and check that it runs without errors

test_labels.py

@@ -7,7 +7,7 @@
 import pytz
 from classes import IssueWithMetrics
 
-from labels import get_average_time_in_labels, get_label_events, get_label_metrics
+from labels import get_stats_time_in_labels, get_label_events, get_label_metrics
 
 
 class TestLabels(unittest.TestCase):
@@ -69,7 +69,7 @@ def test_get_label_metrics_open_issue(self):
 
 
 class TestGetAverageTimeInLabels(unittest.TestCase):
-    """Unit tests for get_average_time_in_labels"""
+    """Unit tests for get_stats_time_in_labels"""
 
     def setUp(self):
         self.issues_with_metrics = MagicMock()
@@ -79,13 +79,14 @@ def setUp(self):
             ),
         ]
 
-    def test_get_average_time_in_labels(self):
-        """Test get_average_time_in_labels"""
+    def test_get_stats_time_in_labels(self):
+        """Test get_stats_time_in_labels"""
         labels = ["bug", "feature"]
-        metrics = get_average_time_in_labels(self.issues_with_metrics, labels)
-        self.assertEqual(len(metrics), 2)
-        self.assertEqual(metrics["bug"], timedelta(days=2))
-        self.assertIsNone(metrics.get("feature"))
+        metrics = get_stats_time_in_labels(self.issues_with_metrics, labels)
+        print(metrics)
+        self.assertEqual(len(metrics['avg']), 2)
+        self.assertEqual(metrics['avg']["bug"], timedelta(days=2))
+        self.assertIsNone(metrics['avg'].get("feature"))
 
 
 if __name__ == "__main__":