Added anomaly detection model details

Author: flanakin

articles/cost-management-billing/understand/analyze-unexpected-charges.md

@@ -52,6 +52,8 @@ Continuing from the previous example of the anomaly labeled **Daily run rate dow
 
 Cost anomalies are evaluated for subscriptions daily and compare the day's total cost to a forecasted total based on the last 60 days to account for common patterns in your recent usage. For example, spikes every Monday. Anomaly detection runs 36 hours after the end of the day (UTC) to ensure a complete data set is available.
 
+The anomaly detection model is a univariate time-series, unsupervised prediction and reconstruction-based model that uses 60 days of historical usage for training, then forecasts expected usage for the day. Anomaly detection forecasting uses a deep learning algorithm called [WaveNet](https://www.deepmind.com/blog/wavenet-a-generative-model-for-raw-audio). Note this is different than the Cost Management forecast. The total normalized usage is determined to be anomalous if it falls outside the expected range based on a predetermined confidence interval.
+
 Anomaly detection is available to every subscription monitored using the cost analysis preview. To enable anomaly detection for your subscriptions, open the cost analysis preview and select your subscription from the scope selector at the top of the page. You'll see a notification informing you that your subscription is onboarded and you'll start to see your anomaly detection status within 24 hours.
 
 ## Manually find unexpected cost changes
@@ -126,4 +128,4 @@ Here's an example email generated for an anomaly alert.
 
 ## Next steps
 
-- Learn about how to [Optimize your cloud investment with Cost Management](../costs/cost-mgt-best-practices.md).
+- Learn about how to [Optimize your cloud investment with Cost Management](../costs/cost-mgt-best-practices.md).