How to plot.

There exist multiple ways to plot using Oríon. We will start with the most convenient one, the experiment object plot accessor.

Accessor experiment.plot

Note

It only supports single experiment plots. It does not support regrets and averages.

It is possible to render most plots directly with the ExperimentClient object. The accessor ExperimentClient.plot can be used to plot the results of the experiment.

from orion.client import get_experiment

# flake8: noqa

# Specify the database where the experiments are stored. We use a local PickleDB here.
storage = dict(type="legacy", database=dict(type="pickleddb", host="../../db.pkl"))

# Load the data for the specified experiment
experiment = get_experiment("2-dim-exp", storage=storage)
fig = experiment.plot.regret()
fig.show()

Module plotting

The plotting module can used as well to plot experiments.

import orion.plotting.base as plot

fig = plot.regret(experiment)
fig.show()

The advantage of the plotting module is that it can render plots using multiple examples as shown in the example below.

fig = plot.regret(experiment)
fig.show()

Command orion plot

Note

The plotting command line utility does not support arguments for now. Contributions are welcome! :)

Note

Only supports single experiment plots. It does not support regrets and averages.

This section refers to the use of the orion plot ... subcommand.

Usage

The arguments expected by the “plot” subcommand are as follows

usage: orion plot [-h] [-n stringID] [-u USER] [-v VERSION] [-c path-to-config]
                  [-t {png,jpg,jpeg,webp,svg,pdf,html,json}]
                  [-o OUTPUT] [--scale SCALE]
                  {lpi,parallel_coordinates,partial_dependencies,regret}

Produce plots for Oríon experiments

positional arguments:
{lpi,parallel_coordinates,partial_dependencies,regret}
                        kind of plot to generate.

optional arguments:
-h, --help            show this help message and exit
-t {png,jpg,jpeg,webp,svg,pdf,html,json}, --type {png,jpg,jpeg,webp,svg,pdf,html,json}
                        type of plot to return. (default: png)
-o OUTPUT, --output OUTPUT
                        path where plot is saved. Will override the `type` argument.
                        (default is {exp.name}-v{exp.version}_{kind}.{type})
--scale SCALE         more pixels, but same proportions of the plot.
                        Scale acts as multiplier on height and width of resulting image.
                        Overrides value of 'scale' in plotly.io.write_image.

Oríon arguments:
These arguments determine orion's behaviour

-n stringID, --name stringID
                        experiment's unique name;
                        (default: None - specified either here or in a config)
-u USER, --user USER  user associated to experiment's unique name;
                        (default: $USER - can be overridden either here or in a config)
-v VERSION, --version VERSION
                        specific version of experiment to fetch; (default: None - latest experiment.)
-c path-to-config, --config path-to-config
                        user provided orion configuration file

More about the arguments

The kind positional argument is the only mandatory argument. Assuming that the user identified the experiment in the usual way (e.g. using --name or a config file), the default behavior is to generate the correct kind of plot, and to save it as a “png” file in the current directory and with a filename automatically formatted as “{experiment.name}_{kind}.png”.

kind

The plotting command requires a kind argument to determine which of the four kinds of plots to generate. The choice is between ‘lpi’, ‘partial_dependencies’, ‘parallel_coordinates’ or ‘regret’.

type

The type is basically the filename extension. This governs more than just the name of the file because it determines the actual format of the output. The default is to give the user a ‘png’ file.

Behind the scenes, plotly generates an initial ‘json’ file, and renders it as an image to be saved in the desired format. With type being ‘json’, that original file is saved without rendering it to an image.

The accepted values of type are ‘png’, ‘jpg’, ‘jpeg’, ‘webp’, ‘svg’, ‘pdf’, ‘html’ and ‘json’.

output

The default value for the output filename is an automatically-generated string formatted as “{experiment.name}_{kind}.{type}”. This also implies that the plot will be saved in the current directory.

When output is specified by the user with a file extension, the type argument will be ignored because the value of type will be instead inferred by the file extension in the output. For example, with --output=../myplot.jpg, the results will be saved in the parent directory and the type will be ‘jpg’.

scale

With certain types of plots, it can be desirable to increase the resolution of the output image in terms of pixel counts (equivalent to dpi). This applies particularly to ‘jpg’ and ‘png’, but it does not affect ‘json’ or ‘html’.

The reference value of scale is 1.0. With --scale 2.0, the height and width are going to be doubled.

Web API

Note

Only supports single experiment plots. It does not support regrets and averages.

The Web API supports queries for single experiment plots. See documentation for all queries. The JSON output is generated automatically according to the Plotly.js schema reference.

Total running time of the script: ( 0 minutes 0.000 seconds)

Gallery generated by Sphinx-Gallery