upmaru/opsmo-crpm

Compute Resource Placement Model (CRPM)

This is a basic Logicstic Regression model that predicts the placement of a compute resource in a compute cluster. The model is trained using the axon library in Elixir. The model is trained on data observed from a compute cluster mixed with some synthetic data.

The synthetic data is used to dictate the expected behaviour of the model to generalize over since we know the expected outcome.

The advantage of solving this problem with an AI model is the following:

1. There is much less code to write and maintain.
2. The model can be retrained with new data when we want to add new feature we simply add columns to the dataset and off we go.

While it is possible to solve this problem with a rule-based system, the amount of code required to maintain and the complexity of the code would be much higher hence this solution.

Source Code

The source code is available on github Opsmo