Cost Anomaly Detection by Monitoring Net Amortized Costs

The question I would like to ask is why you want to use Net Amortized? Cost Anomaly Detection is aiming to identify the spikes when it is happening. Using Net Amortized Cost is sort of defeating the purpose since Amortized Cost is good to gain insights into the effective or normalized daily costs associated with your reservation (RIs or SPs) or subscription portfolio.

While there's already an excellent answer on the other version of this question you asked, it's important to note that Cost Anomaly Detection allows you to submit feedback on the anomalies. This is mentioned briefly in step 8 in the Getting Started documentation for Cost Anomaly Detection.

The feedback you submit on the anomaly indicates whether the anomaly was "accurate" (eg. a real anomaly in usage), a "false positive" (eg. not an anomaly, but rather an expected pattern of usage), or "not an issue" (eg. anomalous usage, but expected - perhaps due to the launch of a new workload).

By providing feedback on the anomalies, Cost Anomaly Detection learns what usage is expected. Over time, this is factored into the daily, weekly, and monthly patterns that Cost Anomaly Detection looks for in your usage.

In your other question and in the comments on this question, you mention spikes due to reservations being detected as anomalies. By providing feedback on these anomalies, in this case using the "false positive" response, the model will learn that reservation spikes are part of the expected pattern. Once Cost Anomaly Detection has enough data & feedback to detect the monthly pattern, Cost Anomaly Detection will recognize the usage as expected rather than anomalous.

Understand your requirement. But if you only want to monitor RI or SP, AWS Cost Anomaly Detection is not the option as Cost Anomaly Detection doesn't monitor any upfront and recurring reserved fee and Savings Plan fees (Reference here). What insights are you trying to gain for RIs or SPs? If you want to monitor coverage and utilization, you can use AWS Budget