An anomaly in OpenSearch is any unusual behavior change in your time-series data. Anomalies can provide valuable insights into your data. For example, for IT infrastructure data, an anomaly in the memory usage metric might help you uncover early signs of a system failure.
It can be challenging to discover anomalies using conventional methods such as creating visualizations and dashboards. You could configure an alert based on a static threshold, but this requires prior domain knowledge and isn't adaptive to data that exhibits organic growth or seasonal behavior.
Anomaly detection automatically detects anomalies in your OpenSearch data in near real-time using the Random Cut Forest (RCF) algorithm. RCF is an unsupervised machine learning algorithm that models a sketch of your incoming data stream to compute an `anomaly grade` and `confidence score` value for each incoming data point. These values are used to differentiate an anomaly from normal variations. For more information about how RCF works, see [Random Cut Forests](https://pdfs.semanticscholar.org/8bba/52e9797f2e2cc9a823dbd12514d02f29c8b9.pdf?_ga=2.56302955.1913766445.1574109076-1059151610.1574109076).
You can pair the anomaly detection plugin with the [alerting plugin]({{site.url}}{{site.baseurl}}/monitoring-plugins/alerting/) to notify you as soon as an anomaly is detected.
To use the anomaly detection plugin, your computer needs to have more than one CPU core.
{: .note }
## Get started with Anomaly Detection
To get started, choose **Anomaly Detection** in OpenSearch Dashboards.
To first test with sample streaming data, choose **Sample Detectors** and try out one of the preconfigured detectors.
### Step 1: Create a detector
A detector is an individual anomaly detection task. You can create multiple detectors, and all the detectors can run simultaneously, with each analyzing data from different sources.
1. Choose **Create Detector**.
1. Enter a name and brief description. Make sure the name is unique and descriptive enough to help you to identify the purpose of the detector.
1. For **Data source**, choose the index you want to use as the data source. You can optionally use index patterns to choose multiple indices.
1. Select the **Timestamp field** in your index.
1. (Optional) For **Data filter**, filter the index you chose as the data source. From the **Filter type** menu, choose **Visual filter**, and then design your filter query by selecting **Fields**, **Operator**, and **Value**, or choose **Custom Expression** and add your own JSON filter query.
1. For **Detector operation settings**, define the **Detector interval**, which is the time interval at which the detector collects data.
- The detector aggregates the data in this interval, then feeds the aggregated result into the anomaly detection model.
The shorter you set this interval, the fewer data points the detector aggregates.
The anomaly detection model uses a shingling process, a technique that uses consecutive data points to create a sample for the model. This process needs a certain number of aggregated data points from contiguous intervals.
- We recommend setting the detector interval based on your actual data. If it's too long it might delay the results, and if it's too short it might miss some data. It also won't have a sufficient number of consecutive data points for the shingle process.
1. (Optional) To add extra processing time for data collection, specify a **Window delay** value. This value tells the detector that the data is not ingested into OpenSearch in real time but with a certain delay.
Set the window delay to shift the detector interval to account for this delay.
- For example, say the detector interval is 10 minutes and data is ingested into your cluster with a general delay of 1 minute.
Assume the detector runs at 2:00. The detector attempts to get the last 10 minutes of data from 1:50 to 2:00, but because of the 1-minute delay, it only gets 9 minutes of data and misses the data from 1:59 to 2:00.
Setting the window delay to 1 minute shifts the interval window to 1:49 - 1:59, so the detector accounts for all 10 minutes of the detector interval time.
1. Choose **Create**.
After you create the detector, the next step is to add features to it.
### Step 2: Add features to your detector
A feature is the field in your index that you want to check for anomalies. A detector can discover anomalies across one or more features. You must choose an aggregation method for each feature: `average()`, `count()`, `sum()`, `min()`, or `max()`. The aggregation method determines what constitutes an anomaly.
For example, if you choose `min()`, the detector focuses on finding anomalies based on the minimum values of your feature. If you choose `average()`, the detector finds anomalies based on the average values of your feature.
A multi-feature model correlates anomalies across all its features. The [curse of dimensionality](https://en.wikipedia.org/wiki/Curse_of_dimensionality) makes it less likely for multi-feature models to identify smaller anomalies as compared to a single-feature model. Adding more features might negatively impact the [precision and recall](https://en.wikipedia.org/wiki/Precision_and_recall) of a model. A higher proportion of noise in your data might further amplify this negative impact. Selecting the optimal feature set is usually an iterative process. We recommend experimenting with a historical detector with different feature sets and checking the precision before moving on to real-time detectors. By default, the maximum number of features for a detector is 5. You can adjust this limit with the `plugins.anomaly_detection.max_anomaly_features` setting.
{: .note }
1. On the **Model configuration** page, enter the **Feature name**.
1. For **Find anomalies based on**, choose the method to find anomalies. For **Field Value** menu, choose the **field** and the **aggregation method**. Or choose **Custom expression**, and add your own JSON aggregation query.
#### (Optional) Set a category field for high cardinality
You can categorize anomalies based on a keyword or IP field type.
The category field categorizes or slices the source time series with a dimension like IP addresses, product IDs, country codes, and so on. This helps to see a granular view of anomalies within each entity of the category field to isolate and debug issues.
To set a category field, choose **Enable a category field** and select a field.
Only a certain number of unique entities are supported in the category field. Use the following equation to calculate the recommended total number of entities supported in a cluster:
```
(data nodes * heap size * anomaly detection maximum memory percentage) / (entity size of a detector)
```
This formula provides a good starting point, but make sure to test with a representative workload.
{: .note }
For example, for a cluster with 3 data nodes, each with 8G of JVM heap size, a maximum memory percentage of 10% (default), and the entity size of the detector as 1MB: the total number of unique entities supported is (8.096 * 10^9 * 0.1 / 1M ) * 3 = 2429.
#### Set a window size
Set the number of aggregation intervals from your data stream to consider in a detection window. It's best to choose this value based on your actual data to see which one leads to the best results for your use case.
Based on experiments performed on a wide variety of one-dimensional data streams, we recommend using a window size between 1 and 16. The default window size is 8. If you set the category field for high cardinality, the default window size is 1.
If you expect missing values in your data or if you want to base the anomalies on the current interval, choose 1. If your data is continuously ingested and you want to base the anomalies on multiple intervals, choose a larger window size.
#### Preview sample anomalies
Preview sample anomalies and adjust the feature settings if needed.
For sample previews, the anomaly detection plugin selects a small number of data samples---for example, one data point every 30 minutes---and uses interpolation to estimate the remaining data points to approximate the actual feature data. It loads this sample dataset into the detector. The detector uses this sample dataset to generate a sample preview of anomaly results.
Examine the sample preview and use it to fine-tune your feature configurations (for example, enable or disable features) to get more accurate results.
1. Choose **Save and start detector**.
1. Choose between automatically starting the detector (recommended) or manually starting the detector at a later time.
### Step 3: Observe the results
Choose the **Anomaly results** tab. You need to wait for some time to see the anomaly results. If the detector interval is 10 minutes, the detector might take more than an hour to start, as it's waiting for sufficient data to generate anomalies.
A shorter interval means the model passes the shingle process more quickly and starts to generate the anomaly results sooner.
Use the [profile detector]({{site.url}}{{site.baseurl}}/monitoring-plugins/ad/api#profile-detector) operation to make sure you have sufficient data points.
If you see the detector pending in "initialization" for longer than a day, aggregate your existing data using the detector interval to check for any missing data points. If you find a lot of missing data points from the aggregated data, consider increasing the detector interval.
Analize anomalies with the following visualizations:
- **Live anomalies** - displays live anomaly results for the last 60 intervals. For example, if the interval is 10, it shows results for the last 600 minutes. The chart refreshes every 30 seconds.
- **Anomaly history** - plots the anomaly grade with the corresponding measure of confidence.
- **Feature breakdown** - plots the features based on the aggregation method. You can vary the date-time range of the detector.
- **Anomaly occurrence** - shows the `Start time`, `End time`, `Data confidence`, and `Anomaly grade` for each detected anomaly.
`Anomaly grade` is a number between 0 and 1 that indicates how anomalous a data point is. An anomaly grade of 0 represents “not an anomaly,” and a non-zero value represents the relative severity of the anomaly.
`Data confidence` is an estimate of the probability that the reported anomaly grade matches the expected anomaly grade. Confidence increases as the model observes more data and learns the data behavior and trends. Note that confidence is distinct from model accuracy.
If you set the category field, you see an additional **Heat map** chart. The heat map correlates results for anomalous entities. This chart is empty until you select an anomalous entity. You also see the anomaly and feature line chart for the time period of the anomaly (`anomaly_grade` > 0).
Choose a filled rectangle to see a more detailed view of the anomaly.
Choose **Set up alerts** and configure a monitor to notify you when anomalies are detected. For steps to create a monitor and set up notifications based on your anomaly detector, see [Monitors]({{site.url}}{{site.baseurl}}/monitoring-plugins/alerting/monitors/).
If you stop or delete a detector, make sure to delete any monitors associated with it.
### Step 5: Adjust the model
To see all the configuration settings for a detector, choose the **Detector configuration** tab.
1. To make any changes to the detector configuration, or fine tune the time interval to minimize any false positives, go to the **Detector configuration** section and choose **Edit**.
- You need to stop the detector to change its configuration. Confirm that you want to stop the detector and proceed.
1. To enable or disable features, in the **Features** section, choose **Edit** and adjust the feature settings as needed. After you make your changes, choose **Save and start detector**.
- Choose between automatically starting the detector (recommended) or manually starting the detector at a later time.
### Step 6: Analyze historical data
Analyzing historical data helps you get familiar with the anomaly detection plugin. You can also evaluate the performance of a detector with historical data to further fine-tune it.
To use a historical detector, you need to specify a date range that has data present in at least 1,000 detection intervals.
{: .note }
1. Choose **Historical detectors** and **Create historical detector**.
1. Enter the **Name** of the detector and a brief **Description**.
1. For **Data source**, choose the index to use as the data source. You can optionally use index patterns to choose multiple indices.
1. For **Time range**, select a time range for historical analysis.
1. For **Detector settings**, choose to use the settings of an existing detector. Or choose the **Timestamp field** in your index, add individual features to the detector, and set the detector interval.
1. (Optional) Choose to run the historical detector automatically after creating it.
1. Choose **Create**.
- You can stop the historical detector even before it completes.
### Step 7: Manage your detectors
To change or delete a detector, go to the **Detector details** page.
1. To make changes to your detector, choose the detector name.
1. Choose **Actions** and **Edit detector**.
- You need to stop the detector to change its configuration. Confirm that you want to stop the detector and proceed.
1. Make your changes and choose **Save changes**.
To delete your detector, choose **Actions** and **Delete detector**. In the pop-up box, type `delete` to confirm and choose **Delete**.
