Add Median and 90th percentile to issue stats.

Currently we are only looking at averages. This adds median and
90th percentile to issues stats. Those are much less sensitive
to outliers. (Think issue cleanup sessions where very old issues
get closed out but then dominate the issue stats at the end of the
month).

labels.py

@@ -3,6 +3,7 @@
 from typing import List
 
 import github3
+import numpy
 import pytz
 
 from classes import IssueWithMetrics
@@ -93,27 +94,34 @@ def get_average_time_in_labels(
     labels: List[str],
 ) -> dict[str, timedelta]:
     """Calculate the average time spent in each label."""
-    average_time_in_labels = {}
-    number_of_issues_in_labels = {}
+    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 in time_in_labels:
+        average_time_in_labels[label] = numpy.average(time_in_labels[label])
+        med_time_in_labels[label] = numpy.median(time_in_labels[label])
+        ninety_percentile_in_labels[label] = numpy.percentile(time_in_labels[label], 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 +170,45 @@ 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:
+      if (stats_time_to_first_response != None):
         file.write(
-            f"| Average time to first response | {average_time_to_first_response} |\n"
+            f"| Time to first response | {stats_time_to_first_response['avg']} | {stats_time_to_first_response['med']} | {stats_time_to_first_response['90p']} |\n"
         )
+      else:
+       file.write(f"| 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 != None):
+        file.write(
+            f"| Time to close | {stats_time_to_close['avg']} | {stats_time_to_close['med']} | {stats_time_to_close['90p']} |\n"
+        )
+       else:
+        file.write(f"| 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 != None):
+        file.write(
+            f"| Time to answer | {stats_time_to_answer['avg']} | {stats_time_to_answer['med']} | {stats_time_to_answer['90p']} |\n"
+        )
+       else:
+        file.write(f"| 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} | {stats_time_in_labels['avg'][label]} | {stats_time_in_labels['med'][label]} | {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")

test_labels.py

@@ -83,9 +83,10 @@ def test_get_average_time_in_labels(self):
         """Test get_average_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"))
+        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__":

test_markdown_writer.py

@@ -17,7 +17,7 @@
 
 class TestWriteToMarkdown(unittest.TestCase):
     """Test the write_to_markdown function."""
-
+    maxDiff = None
     def test_write_to_markdown(self):
         """Test that write_to_markdown writes the correct markdown file.
 
@@ -48,20 +48,32 @@ def test_write_to_markdown(self):
                 {"bug": timedelta(days=2)},
             ),
         ]
-        average_time_to_first_response = timedelta(days=2)
-        average_time_to_close = timedelta(days=3)
-        average_time_to_answer = timedelta(days=4)
-        average_time_in_labels = {"bug": "1 day, 12:00:00"}
+        time_to_first_response = {
+                'avg': timedelta(days=2),
+                'med': timedelta(days=2),
+                '90p': timedelta(days=2)}
+        time_to_close = {
+                'avg': timedelta(days=3),
+                'med': timedelta(days=3),
+                '90p': timedelta(days=3)}
+        time_to_answer = {
+                'avg': timedelta(days=4),
+                'med': timedelta(days=4),
+                '90p': timedelta(days=4)}
+        time_in_labels = {
+                'avg': {"bug": "1 day, 12:00:00"},
+                'med': {"bug": "1 day, 12:00:00"},
+                '90p': {"bug": "1 day, 12:00:00"}}
         num_issues_opened = 2
         num_issues_closed = 1
 
         # Call the function
         write_to_markdown(
             issues_with_metrics=issues_with_metrics,
-            average_time_to_first_response=average_time_to_first_response,
-            average_time_to_close=average_time_to_close,
-            average_time_to_answer=average_time_to_answer,
-            average_time_in_labels=average_time_in_labels,
+            average_time_to_first_response=time_to_first_response,
+            average_time_to_close=time_to_close,
+            average_time_to_answer=time_to_answer,
+            average_time_in_labels=time_in_labels,
             num_issues_opened=num_issues_opened,
             num_issues_closed=num_issues_closed,
             labels=["bug"],
@@ -73,12 +85,12 @@ def test_write_to_markdown(self):
             content = file.read()
         expected_content = (
             "# Issue Metrics\n\n"
-            "| Metric | Value |\n"
-            "| --- | ---: |\n"
-            "| Average time to first response | 2 days, 0:00:00 |\n"
-            "| Average time to close | 3 days, 0:00:00 |\n"
-            "| Average time to answer | 4 days, 0:00:00 |\n"
-            "| Average time spent in bug | 1 day, 12:00:00 |\n"
+            "| Metric | Average | Median | 90th percentile |\n"
+            "| --- | --- | --- | ---: |\n"
+            "| Time to first response | 2 days, 0:00:00 | 2 days, 0:00:00 | 2 days, 0:00:00 |\n"
+            "| Time to close | 3 days, 0:00:00 | 3 days, 0:00:00 | 3 days, 0:00:00 |\n"
+            "| Time to answer | 4 days, 0:00:00 | 4 days, 0:00:00 | 4 days, 0:00:00 |\n"
+            "| Time spent in bug | 1 day, 12:00:00 | 1 day, 12:00:00 | 1 day, 12:00:00 |\n"
             "| Number of items that remain open | 2 |\n"
             "| Number of items closed | 1 |\n"
             "| Total number of items created | 2 |\n\n"
@@ -125,10 +137,22 @@ def test_write_to_markdown_with_vertical_bar_in_title(self):
                 {"bug": timedelta(days=2)},
             ),
         ]
-        average_time_to_first_response = timedelta(days=2)
-        average_time_to_close = timedelta(days=3)
-        average_time_to_answer = timedelta(days=4)
-        average_time_in_labels = {"bug": "1 day, 12:00:00"}
+        average_time_to_first_response = {
+             'avg' : timedelta(days=2),
+             'med' : timedelta(days=2),
+             '90p' : timedelta(days=2)}
+        average_time_to_close = {
+             'avg' : timedelta(days=3),
+             'med' : timedelta(days=3),
+             '90p' : timedelta(days=3)}
+        average_time_to_answer = {
+             'avg' : timedelta(days=4),
+             'med' : timedelta(days=4),
+             '90p' : timedelta(days=4)}
+        average_time_in_labels = {
+             'avg' : {"bug": "1 day, 12:00:00"},
+             'med' : {"bug": "1 day, 12:00:00"},
+             '90p' : {"bug": "1 day, 12:00:00"}}
         num_issues_opened = 2
         num_issues_closed = 1
 
@@ -149,12 +173,12 @@ def test_write_to_markdown_with_vertical_bar_in_title(self):
             content = file.read()
         expected_content = (
             "# Issue Metrics\n\n"
-            "| Metric | Value |\n"
-            "| --- | ---: |\n"
-            "| Average time to first response | 2 days, 0:00:00 |\n"
-            "| Average time to close | 3 days, 0:00:00 |\n"
-            "| Average time to answer | 4 days, 0:00:00 |\n"
-            "| Average time spent in bug | 1 day, 12:00:00 |\n"
+            "| Metric | Average | Median | 90th percentile |\n"
+            "| --- | --- | --- | ---: |\n"
+            "| Time to first response | 2 days, 0:00:00 | 2 days, 0:00:00 | 2 days, 0:00:00 |\n"
+            "| Time to close | 3 days, 0:00:00 | 3 days, 0:00:00 | 3 days, 0:00:00 |\n"
+            "| Time to answer | 4 days, 0:00:00 | 4 days, 0:00:00 | 4 days, 0:00:00 |\n"
+            "| Time spent in bug | 1 day, 12:00:00 | 1 day, 12:00:00 | 1 day, 12:00:00 |\n"
             "| Number of items that remain open | 2 |\n"
             "| Number of items closed | 1 |\n"
             "| Total number of items created | 2 |\n\n"
@@ -259,8 +283,8 @@ def test_writes_markdown_file_with_non_hidden_columns_only(self):
             content = file.read()
         expected_content = (
             "# Issue Metrics\n\n"
-            "| Metric | Value |\n"
-            "| --- | ---: |\n"
+            "| Metric | Average | Median | 90th percentile |\n"
+            "| --- | --- | --- | ---: |\n"
             "| Number of items that remain open | 2 |\n"
             "| Number of items closed | 1 |\n"
             "| Total number of items created | 2 |\n\n"

test_time_to_answer.py

@@ -59,7 +59,7 @@ def test_returns_average_time_to_answer(self):
         ]
 
         # Act
-        result = get_average_time_to_answer(issues_with_metrics)
+        result = get_average_time_to_answer(issues_with_metrics)['avg']
 
         # Assert
         self.assertEqual(result, timedelta(seconds=20))

test_time_to_close.py

@@ -44,7 +44,7 @@ def test_get_average_time_to_close(self):
         ]
 
         # Call the function and check the result
-        result = get_average_time_to_close(issues_with_metrics)
+        result = get_average_time_to_close(issues_with_metrics)['avg']
         expected_result = timedelta(days=3)
         self.assertEqual(result, expected_result)
 

test_time_to_first_response.py

@@ -334,7 +334,7 @@ def test_get_average_time_to_first_response(self):
         ]
 
         # Call the function and check the result
-        result = get_average_time_to_first_response(issues_with_metrics)
+        result = get_average_time_to_first_response(issues_with_metrics)['avg']
         expected_result = timedelta(days=1.5)
         self.assertEqual(result, expected_result)
 

time_to_answer.py

@@ -44,12 +44,19 @@ def get_average_time_to_answer(
     num_issues_with_time_to_answer = len(issues_with_time_to_answer)
     if num_issues_with_time_to_answer > 0:
         average_time_to_answer = numpy.average(answer_times)
+        med_time_to_answer = numpy.median(answer_times)
+        ninety_percentile_time_to_answer = numpy.percentile(answer_times, 90, axis=0)
     else:
         return None
 
+    stats = {
+                 'avg': timedelta(seconds=average_time_to_answer),
+                 'med': timedelta(seconds=med_time_to_answer),
+                 '90p': timedelta(seconds=ninety_percentile_time_to_answer)}
+
     # Print the average time to answer converting seconds to a readable time format
     print(f"Average time to answer: {timedelta(seconds=average_time_to_answer)}")
-    return timedelta(seconds=average_time_to_answer)
+    return stats
 
 
 def measure_time_to_answer(discussion: dict) -> Union[timedelta, None]:

time_to_close.py

@@ -85,9 +85,16 @@ def get_average_time_to_close(
     num_issues_with_time_to_close = len(issues_with_time_to_close)
     if num_issues_with_time_to_close > 0 and total_time_to_close is not None:
         average_time_to_close = numpy.average(close_times)
+        med_time_to_close = numpy.median(close_times)
+        ninety_percentile_time_to_close = numpy.percentile(close_times, 90, axis=0)
     else:
         return None
 
+    stats = {
+                 'avg': timedelta(seconds=average_time_to_close),
+                 'med': timedelta(seconds=med_time_to_close),
+                 '90p': timedelta(seconds=ninety_percentile_time_to_close)}
+
     # Print the average time to close converting seconds to a readable time format
-    print(f"Average time to close: {timedelta(seconds=average_time_to_close)}")
-    return timedelta(seconds=average_time_to_close)
+    print(f"Time to close: {timedelta(seconds=average_time_to_close)}")
+    return stats

time_to_first_response.py

@@ -146,10 +146,17 @@ def get_average_time_to_first_response(
         return None
 
     average_seconds_to_first_response = numpy.average(response_times)
+    med_seconds_to_first_response = numpy.median(response_times)
+    ninety_percentile_seconds_to_first_response = numpy.percentile(response_times, 90, axis=0)
+
+    stats = {
+                 'avg': timedelta(seconds=average_seconds_to_first_response),
+                 'med': timedelta(seconds=med_seconds_to_first_response),
+                 '90p': timedelta(seconds=ninety_percentile_seconds_to_first_response)}
 
     # Print the average time to first response converting seconds to a readable time format
     print(
         f"Average time to first response: {timedelta(seconds=average_seconds_to_first_response)}"
     )
 
-    return timedelta(seconds=average_seconds_to_first_response)
+    return stats