remove CSV onboarding for Metrics Advisor

Author: jr-MS

articles/applied-ai-services/metrics-advisor/data-feeds-from-different-sources.md

@@ -533,16 +533,6 @@ For more information, refer to the [tutorial on writing a valid query](tutorials
     ```
     For more information, refer to the [tutorial on writing a valid query](tutorials/write-a-valid-query.md).
 
-## <span id="csv">Local files (CSV)</span>
-
-> [!NOTE]
-> This feature is only used for quick system evaluation focusing on anomaly detection. It only accepts static data from a local CSV, and performs anomaly detection on single time series data. For analyzing multi-dimensional metrics, including real-time data ingestion, anomaly notification, root cause analysis, and cross-metric incident analysis, use other supported data sources.
-
-**Requirements on data in CSV:**
-- Have at least one column, which represents measurements to be analyzed. For a better and quicker user experience, try a CSV file that contains two columns: a timestamp column and a metric column. The timestamp format should be as follows: `2021-03-30T00:00:00Z`, and the `seconds` part is best to be `:00Z`. The time granularity between every record should be the same.
-- The timestamp column is optional. If there's no timestamp, Metrics Advisor will use timestamp starting from today (`00:00:00` Coordinated Universal Time). The service maps each measure in the row at a one-hour interval.
-- There is no re-ordering or gap-filling happening during data ingestion. Make sure that your data in the CSV file is ordered by the timestamp ordering **ascending (ASC)**.
- 
 ## Next steps
 
 * While you're waiting for your metric data to be ingested into the system, read about [how to manage data feed configurations](how-tos/manage-data-feeds.md).

articles/cognitive-services/Anomaly-Detector/overview-multivariate.md

@@ -38,7 +38,7 @@ Basically, Multivariate Anomaly Detector includes three main steps, **data prepa
 For data preparation, you should prepare two parts of data, **training data** and **inference data**. As for training data, you should upload your data to Blob Storage and generate an SAS url which will be used in training API. As for inference data, you could either use the same data format as training data, or send the data into API header which will be a json format. This depends on what API you would use in inference process.
 
 ### Training
-When training a model, you should call an asynchronized API on your training data, which means you won't get the model status immediately after calling this API, you should request another API to get the model status.
+When training a model, you should call an asynchronized API on your training data, which means you won't get the model status immediately after calling this API, you should requessf jjsfd jijisjist another API to get the model status.
 
 ### Inference
 In inference process, you have two options to choose, asynchroized API or synchronized API. If you would like to do a batch validation, you are suggested to use asynchronized API. If you want to do streaming in a short granularity and get the inference result immediately after each API request, you are suggested to use synchronized API.