Skip to main content

Polaris

Project description

Polaris

Polaris logo

pipeline status coverage report

Python3 tool to analyze a satellite set of telemetry to understand links/dependencies among different subsystems. The telemetry is currently retrieved from the SatNOGS Network.

If you want to know more:

If you are a developer, and want to contribute to Polaris check out CONTRIBUTE.md.

Project structure

contrib/               - code that is not directly dependent on Polaris, but is used in the project
docs/                  - Some documentation on the project (though more is in the wiki)
polaris/               - Project source code
    common/            - Modules common to all of Polaris
    fetch/             - Module to fetch and prepare data for the analysis
    viz/               - Module to visualize the analysis results
    learn/             - Module to perform the data analysis
    batch/             - Module to perform batch operations
    convert/           - Module to convert graph output from learn to other file formats
    polaris.py         - Polaris entry point

tests/                 - Project unit tests
playground/            - Exploratory tests

Installation

NOTE: Until https://gitlab.com/librespacefoundation/polaris/polaris/-/issues/124 is resolved, Polaris will not work with Python 3.9. Python 3.8 is known to work, and earlier versions of Python 3.x may work as well. Check your operating system for details, but usually you can run (or install) the correct version by invoking python3.8 instead of just python3 or python.

pip3 install polaris-ml

We recommend to install it inside a Python virtual environment:

# Create the virtual env
$ python3 -m venv .venv

# Activate it
$ source .venv/bin/activate

# Upgrade Pip before installing Polaris
$ (.venv) pip install --upgrade pip

# Install Polaris from Pypi
$ (.venv) pip install polaris-ml

Note: If you run into problems installing Polaris via pip, make sure you've upgraded pip itself and are using a clean, new, separate virtual environment -- this solves most problems.

Running the code

$ (.venv) polaris --help
Usage: polaris [OPTIONS] COMMAND [ARGS]...

  Tool for analyzing satellite telemetry

Options:
  --version   Show the version and exit.
  --help      Show this message and exit.

Commands:
  batch     Run polaris commands in batch mode
  convert   Convert polaris graph file (supported formats: gexf)
  fetch     Download data set(s)
  learn     Analyze data
  viz       Displaying results

# To fetch and decode data from the SatNOGS network and space weather sources, run:
$ (.venv) polaris fetch -s 2019-08-10 -e 2019-10-5 --cache_dir /tmp/LightSail_2 LightSail-2 /tmp/normalized_frames.json
# Note: this may take some time.

# If the normalizer for your satellite does not exist, you can run polaris fetch
# with the --skip_normalizer flag. The result with and without the normalizer
# (without/with --skip-normalizer) are bound to be slightly different.
# The normalizer is mainly present to give you, the satellite operator, an
# intuitive (SI) value for fields (instead of arbitrarily scaled/shifted
# values). It is easy to create and you can get started with the process using
# the snippet at https://gitlab.com/librespacefoundation/polaris/polaris/-/snippets/2006696
$ (.venv) polaris fetch -s 2019-08-10 -e 2019-10-5 --cache_dir /tmp/LightSail_2 --skip_normalizer LightSail-2 /tmp/normalized_frames.json

# Data will be saved at /tmp/normalized_frames.json
$ (.venv) head /tmp/normalized_frames.json
[
    {
        "time": "2019-09-12 08:14:42",
        "measurement": "",
        "tags": {
            "satellite": "",
            "decoder": "Lightsail2",
            "station": "",
            "observer": "",
            "source": "",
[...]


# To learn from that data, run:
$ (.venv) polaris learn -g /tmp/new_graph.json /tmp/normalized_frames.json
# Note: depending on your hardware, this may take some time.

# Note: `polaris learn` uses your dedicated (CUDA enabled) GPU by default
#       to suppress this behaviour, you can utilise the --force-cpu flag.
$ (.venv) polaris learn -g /tmp/new_graph.json /tmp/normalized_frames.json --force_cpu

# To see a visualization of these results, run:
$ (.venv) polaris viz /tmp/new_graph.json
# Then visit http://localhost:8080 in your browser

Configuring Polaris

It is possible to override the default parameters used in the ai processes of Polaris by your own using configuration files.

  • Learn cross correlation process (generating graph) :
{
  "use_gridsearch": false,
  "random_state": 43,
  "test_size": 0.2,
  "gridsearch_scoring": "neg_mean_squared_error",
  "gridsearch_n_splits": 6,
  "dataset_cleaning_params": {
    "col_max_na_percentage": 100,
    "row_max_na_percentage": 100
    },
  "model_cpu_params": {
    "objective": "reg:squarederror",
    "n_estimators": 81,
    "learning_rate": 0.1,
    "n_jobs": 1,
    "predictor": "cpu_predictor",
    "tree_method": "auto",
    "max_depth": 8
    },
  "model_params": {
    "objective": "reg:squarederror",
    "n_estimators": 80,
    "learning_rate": 0.1,
    "n_jobs": 1,
    "max_depth": 8
    }
}

To use it, add the -l or learn_config_file command line parameter when calling learn:

$ polaris learn -g /tmp/graph.json /tmp/normalized_frames -l ../xcorr_cfg.json

## Batch operations

Batch operations allow automation of repeated steps.  For example:

- running `polaris fetch` so that it fetches the latest data for a particular satellite, then running `polaris learn` to update the model

- running `polaris fetch`, `polaris learn` and `polaris viz` as part of an integration test

The `polaris batch` command is controlled by a JSON configuration file; an example can be found at `polaris/common/polaris_config.json.EXAMPLE`.

```bash
$ (.venv) polaris batch --config_file polaris/common/polaris_config.json.EXAMPLE

InfluxDB

With the addition of space weather recently, influxdb support has been added to Polaris. To create the required docker-compose.yml file and start and stop the docker container, run:

$ python
>>> from vinvelivaanilai.storage import common

# To create the path
>>> common.create_docker_compose("/path/to/docker-compose.yml", "/path/to/storage")

# To start influxdb
>>> common.start_docker_compose("/path/to/docker-compose.yml")

# To stop influxdb
>>> common.stop_docker_compose("/path/to/docker-compose.yml")

To store in and fetch from influxdb use the flags --store_in_influxdb and --fetch_from_influxdb respectively.

polaris fetch -s 2019-08-10 -e 2019-10-5 --cache_dir /tmp/LightSail_2 LightSail-2 /tmp/normalized_frames.json --store_in_influxdb
$ polaris fetch -s 2019-08-10 -e 2019-10-5 --cache_dir /tmp/LightSail_2 LightSail-2 /tmp/normalized_frames.json --fetch_from_influxdb

MLflow

Installing Polaris will install MLflow as a dependency. At this time Polaris is using MLflow during the cross check dependencies process, and the database is stored in the current working directory under the mlruns folder.

To view the logs in MLflow, run this command in the directory that holds the mlruns folder (by default, this is the project root directory):

mlflow ui

This command will start the tracking ui server at http://localhost:5000.

Working on documentation

Documentation is hosted on readthedocs.io. We use the Myst parser, and write documentation in Markdown.

To work on documentation, install the docs dependencies like so:

# Yes, with the square brackets:
pip install -e .[docs]

Documentation files are in the docs/ directory. To build the HTML files, run:

cd docs/
make html

Generated files will be in the docs/_build directory, and can be viewed with your favourite browser.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for polaris-ml, version 0.13.0
Filename, size File type Python version Upload date Hashes
Filename, size polaris_ml-0.13.0-py2.py3-none-any.whl (130.6 kB) File type Wheel Python version py2.py3 Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page