C-Dash: Fix anomaly detection for the news

Signed-off-by: Tibor Frank <tifrank@cisco.com>
Change-Id: I63e280e15e82583b65691be17a07208d0ab788ce

5 files changed, 80 insertions, 59 deletions

diff --git a/csit.infra.dash/app/cdash/data/data.yaml b/csit.infra.dash/app/cdash/data/data.yaml
index 8beee0bacc..720109b552 100644
--- a/csit.infra.dash/app/cdash/data/data.yaml
+++ b/csit.infra.dash/app/cdash/data/data.yaml
@@ -240,7 +240,6 @@
     # - result_latency_value
     - start_time
     - passed
-    - telemetry
     - test_id
     - version
 - data_type: coverage
diff --git a/csit.infra.dash/app/cdash/news/layout.py b/csit.infra.dash/app/cdash/news/layout.py
index da36b1430c..d8ad92a1db 100644
--- a/csit.infra.dash/app/cdash/news/layout.py
+++ b/csit.infra.dash/app/cdash/news/layout.py
@@ -24,7 +24,8 @@ from dash import callback_context
 from dash import Input, Output, State
 
 from ..utils.constants import Constants as C
-from ..utils.utils import classify_anomalies, gen_new_url
+from ..utils.utils import gen_new_url
+from ..utils.anomalies import classify_anomalies
 from ..utils.url_processing import url_decode
 from .tables import table_summary
 
@@ -132,15 +133,17 @@ class Layout:
 
             tests = df_job["test_id"].unique()
             for test in tests:
-                tst_data = df_job.loc[df_job["test_id"] == test].sort_values(
-                    by="start_time", ignore_index=True)
-                x_axis = tst_data["start_time"].tolist()
+                tst_data = df_job.loc[(
+                    (df_job["test_id"] == test) &
+                    (df_job["passed"] == True)
+                )].sort_values(by="start_time", ignore_index=True)
                 if "-ndrpdr" in test:
                     tst_data = tst_data.dropna(
                         subset=["result_pdr_lower_rate_value", ]
                     )
                     if tst_data.empty:
                         continue
+                    x_axis = tst_data["start_time"].tolist()
                     try:
                         anomalies, _, _ = classify_anomalies({
                             k: v for k, v in zip(
@@ -185,6 +188,7 @@ class Layout:
                     )
                     if tst_data.empty:
                         continue
+                    x_axis = tst_data["start_time"].tolist()
                     try:
                         anomalies, _, _ = classify_anomalies({
                             k: v for k, v in zip(
diff --git a/csit.infra.dash/app/cdash/trending/graphs.py b/csit.infra.dash/app/cdash/trending/graphs.py
index fc26f8bd79..ba94eefeed 100644
--- a/csit.infra.dash/app/cdash/trending/graphs.py
+++ b/csit.infra.dash/app/cdash/trending/graphs.py
@@ -18,7 +18,8 @@ import plotly.graph_objects as go
 import pandas as pd
 
 from ..utils.constants import Constants as C
-from ..utils.utils import classify_anomalies, get_color, get_hdrh_latencies
+from ..utils.utils import get_color, get_hdrh_latencies
+from ..utils.anomalies import classify_anomalies
 
 
 def select_trending_data(data: pd.DataFrame, itm: dict) -> pd.DataFrame:
diff --git a/csit.infra.dash/app/cdash/utils/anomalies.py b/csit.infra.dash/app/cdash/utils/anomalies.py
new file mode 100644
index 0000000000..9a7b232fda
--- /dev/null
+++ b/csit.infra.dash/app/cdash/utils/anomalies.py
@@ -0,0 +1,69 @@
+# Copyright (c) 2023 Cisco and/or its affiliates.
+# 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.
+
+"""Functions used by Dash applications to detect anomalies.
+"""
+
+from numpy import isnan
+
+from ..jumpavg import classify
+
+
+def classify_anomalies(data):
+    """Process the data and return anomalies and trending values.
+
+    Gather data into groups with average as trend value.
+    Decorate values within groups to be normal,
+    the first value of changed average as a regression, or a progression.
+
+    :param data: Full data set with unavailable samples replaced by nan.
+    :type data: OrderedDict
+    :returns: Classification and trend values
+    :rtype: 3-tuple, list of strings, list of floats and list of floats
+    """
+    # NaN means something went wrong.
+    # Use 0.0 to cause that being reported as a severe regression.
+    bare_data = [0.0 if isnan(sample) else sample for sample in data.values()]
+    # TODO: Make BitCountingGroupList a subclass of list again?
+    group_list = classify(bare_data).group_list
+    group_list.reverse()  # Just to use .pop() for FIFO.
+    classification = list()
+    avgs = list()
+    stdevs = list()
+    active_group = None
+    values_left = 0
+    avg = 0.0
+    stdv = 0.0
+    for sample in data.values():
+        if isnan(sample):
+            classification.append("outlier")
+            avgs.append(sample)
+            stdevs.append(sample)
+            continue
+        if values_left < 1 or active_group is None:
+            values_left = 0
+            while values_left < 1:  # Ignore empty groups (should not happen).
+                active_group = group_list.pop()
+                values_left = len(active_group.run_list)
+            avg = active_group.stats.avg
+            stdv = active_group.stats.stdev
+            classification.append(active_group.comment)
+            avgs.append(avg)
+            stdevs.append(stdv)
+            values_left -= 1
+            continue
+        classification.append("normal")
+        avgs.append(avg)
+        stdevs.append(stdv)
+        values_left -= 1
+    return classification, avgs, stdevs
diff --git a/csit.infra.dash/app/cdash/utils/utils.py b/csit.infra.dash/app/cdash/utils/utils.py
index d9347b1c13..29bee3d039 100644
--- a/csit.infra.dash/app/cdash/utils/utils.py
+++ b/csit.infra.dash/app/cdash/utils/utils.py
@@ -11,7 +11,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-"""Function used by Dash applications.
+"""Functions used by Dash applications.
 """
 
 import pandas as pd
@@ -22,65 +22,13 @@ import hdrh.histogram
 import hdrh.codec
 
 from math import sqrt
-from numpy import isnan
 from dash import dcc
 from datetime import datetime
 
-from ..jumpavg import classify
 from ..utils.constants import Constants as C
 from ..utils.url_processing import url_encode
 
 
-def classify_anomalies(data):
-    """Process the data and return anomalies and trending values.
-
-    Gather data into groups with average as trend value.
-    Decorate values within groups to be normal,
-    the first value of changed average as a regression, or a progression.
-
-    :param data: Full data set with unavailable samples replaced by nan.
-    :type data: OrderedDict
-    :returns: Classification and trend values
-    :rtype: 3-tuple, list of strings, list of floats and list of floats
-    """
-    # NaN means something went wrong.
-    # Use 0.0 to cause that being reported as a severe regression.
-    bare_data = [0.0 if isnan(sample) else sample for sample in data.values()]
-    # TODO: Make BitCountingGroupList a subclass of list again?
-    group_list = classify(bare_data).group_list
-    group_list.reverse()  # Just to use .pop() for FIFO.
-    classification = list()
-    avgs = list()
-    stdevs = list()
-    active_group = None
-    values_left = 0
-    avg = 0.0
-    stdv = 0.0
-    for sample in data.values():
-        if isnan(sample):
-            classification.append("outlier")
-            avgs.append(sample)
-            stdevs.append(sample)
-            continue
-        if values_left < 1 or active_group is None:
-            values_left = 0
-            while values_left < 1:  # Ignore empty groups (should not happen).
-                active_group = group_list.pop()
-                values_left = len(active_group.run_list)
-            avg = active_group.stats.avg
-            stdv = active_group.stats.stdev
-            classification.append(active_group.comment)
-            avgs.append(avg)
-            stdevs.append(stdv)
-            values_left -= 1
-            continue
-        classification.append("normal")
-        avgs.append(avg)
-        stdevs.append(stdv)
-        values_left -= 1
-    return classification, avgs, stdevs
-
-
 def get_color(idx: int) -> str:
     """Returns a color from the list defined in Constants.PLOT_COLORS defined by
     its index.