Fix mypy and fixed timeseries

getsentry · Nov 4, 2024 · 7ae6de5 · 7ae6de5
1 parent 5339865
commit 7ae6de5
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 7 deletions.
diff --git a/src/seer/anomaly_detection/anomaly_detection.py b/src/seer/anomaly_detection/anomaly_detection.py
@@ -147,17 +147,16 @@ def _online_detect(
             history_mp=anomalies.matrix_profile,
             window_size=anomalies.window_size,
             history_flags=anomalies.flags,
-            anomaly_algo_data=anomalies.get_anomaly_algo_data(len(historic.timeseries.timestamps))[
-                -5:  # TODO: This slice should be based on the time period
-            ],
+            anomaly_algo_data=anomalies.get_anomaly_algo_data(len(historic.timeseries.timestamps)),
         )
         streamed_anomalies = stream_detector.detect(
             convert_external_ts_to_internal(ts_external), config
         )
 
         # Get current point's anomaly data and track original flag
         curr_algo_data = streamed_anomalies.get_anomaly_algo_data(len(ts_external))[0]
-        curr_algo_data["original_flag"] = streamed_anomalies.original_flags[-1]
+        if streamed_anomalies.original_flags and curr_algo_data is not None:
+            curr_algo_data["original_flag"] = streamed_anomalies.original_flags[-1]
 
         # Save new data point
         alert_data_accessor.save_timepoint(
@@ -240,6 +239,8 @@ def _combo_detect(
             history_values=historic.values,
             history_mp=anomalies.matrix_profile,
             window_size=anomalies.window_size,
+            history_flags=anomalies.flags,
+            anomaly_algo_data=anomalies.get_anomaly_algo_data(len(historic.timestamps)),
         )
         streamed_anomalies = stream_detector.detect(
             convert_external_ts_to_internal(ts_external), config

diff --git a/src/seer/anomaly_detection/detectors/anomaly_detectors.py b/src/seer/anomaly_detection/detectors/anomaly_detectors.py
@@ -155,6 +155,11 @@ class MPStreamAnomalyDetector(AnomalyDetector):
         arbitrary_types_allowed=True,
     )
 
+    stream_history_size: dict[int, int] = Field(
+        default={5: 19, 15: 11, 30: 7, 60: 5},
+        description="History size for stream smoothing based on the function smooth_size = floor(43 / sqrt(time_period))",
+    )
+
     @inject
     @sentry_sdk.trace
     def detect(
@@ -215,9 +220,13 @@ def detect(
                 if flags_and_scores is None:
                     raise ServerError("Failed to score the matrix profile distance")
 
+                anomaly_algo_data_to_use = self.anomaly_algo_data[
+                    self.stream_history_size[config.time_period]
+                ]
+
                 # The original flags are the flags of the previous points
                 past_original_flags = [
-                    algo_data["original_flag"] for algo_data in self.anomaly_algo_data
+                    algo_data["original_flag"] for algo_data in anomaly_algo_data_to_use
                 ]
 
                 self.anomaly_algo_data.append({"original_flag": flags_and_scores.flags[-1]})

diff --git a/src/seer/anomaly_detection/detectors/window_size_selectors.py b/src/seer/anomaly_detection/detectors/window_size_selectors.py
@@ -149,5 +149,4 @@ def optimal_window_size(self, time_series: npt.NDArray[np.float64]) -> int:
             sentry_sdk.set_tag(AnomalyDetectionTags.WINDOW_SEARCH_FAILED, 1)
             return 3
         sentry_sdk.set_tag(AnomalyDetectionTags.WINDOW_SEARCH_FAILED, 0)
-        # return window_size
-        return 10
+        return window_size