0
$\begingroup$

I'm confused on how to go about an issue.

I'm trying to implement an unsupervised model, using a dataset that is essentially a log file. This dataset contains a variety of features, but most importantly, it contains a column that identifies the individual that "created" the log entry. It also contains features such as a timestamp, and more regarding that specific event.

I want to be able to detect if a sample is an anomaly for that specific individual. So while the dataset might contain any number of individuals, if person A does something, I want to compare that to person A:s previous events, and not against the entire dataset of samples.

The answer might be obvious, but I'm just stumped.

While I could simply just create a model for each person, that feels kinda dumb. However, that might be the simplest option, by training a separate model for each person and then just loading the appropriate model when a sample occurs.

I don't know, I'm stuck. Please help.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

For a strategy that treats each "group" (in your case a person) as independent, the most straightforward is indeed to create one model per group, each with its own set of parameters.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.