Taking a look at the discription of both kinds, orientation seems the same. while it isn't the same I can't figure the difference. In a Heuristic scan it looks for suspicious or malicious behaviors in a file, Anomaly analysis looks for anomalies in a file and its structure.
Search for anomaly is look for things that aren't bad per se, but usually isn't common. For instance, a freshly registered domain being accessed late night, and receiving lots of traffic. Or your HR computers accessing Github, or executing PowerShell scripts.
Those things are not wrong, they are out of place. Accessing Github is not wrong. But it's strange for certain group of users. In this case, an heuristic scanner would not find anything strange.
On the other hand, an anomaly detector needs a baseline. If you don't have a good baseline, you will either get a lot of false positives, and end up shutting down the scanner, or a lot of false negatives and will not be of any use.