Error Analysis

In this small project we will:

1. start with current errors from the spack monitor API
2. generate DBSTREAM clusters using river to describe them
3. save output of cluster assignments to file
4. generate another classification model to associate with features

Right now we have a basic set of errors, and we will need to get specs (the features) in this repo too. @vsoch will wait to see a spec from Dinos set and then ensure the spack monitor ones are exported to be the same.

Usage

$ python -m venv env
$ source env/bin/activate
$ conda install river
$ pip install -r requirements.txt

Important @vsoch fixed some bugs with dbstream so you'll need to install river from GitHub

$ git clone https://github.com/online-ml/river
$ cd river
$ pip install -e .

For now we are using already downloaded data from spack monitor instances, provided in the repository (minus the specs so far!)

Online ML

The script 2.online-ml.py will generate the clusters in data/clusters. The default number of iterations is 5 but you can specify a custom value:

$ python 2.online-ml.py --iter 5

License

Spack is distributed under the terms of both the MIT license and the Apache License (Version 2.0). Users may choose either license, at their option.

All new contributions must be made under both the MIT and Apache-2.0 licenses.

See LICENSE-MIT, LICENSE-APACHE, COPYRIGHT, and NOTICE for details.

SPDX-License-Identifier: (Apache-2.0 OR MIT)

LLNL-CODE-811652