autograde

Autograding package for EA and other classes

Earth Py

A package built to support python teaching in the Earth Lab earth analytics program at University of Colorado, Boulder.

Install

To install, use pip. --upgrade is optional but it ensures that the package overwrites when you install and you have the current version. If you don't have the package yet you can still use the --upgrade argument.

pip install --upgrade git+https://github.com/earthlab/autograde.git

Then import it into python.

import autograde as gr

Background

This library was developed to simplify the autograding process of Matplotlib plots. Visually similar plots can be created in a variety of ways and hold different metadata. Our goal is to abstract away these differences by creating a simple way to test student plots.

Beyond that, we have noticed common groupings of assertions for specific plot types. PlotBasicSuiteobjects have been created to avoid repetition in writing out assertions, and return a TestSuite instead. To run the suite after it has been created, use a unittest text runner.

Examples

2D plot with x-axis label containing "x" and y-axis label containing "y" and "data"

from autograde.cases import PlotBasicSuite
import pandas as pd
import unittest

axis = plt.gca()
data = pd.DataFrame(data={“x”:xvals, “y”:yvals})
suite = PlotBasicSuite(ax=axis, data_exp=data, xcol=”x”, ycol=”y”)
xlabel_contains=[“x”], ylabel_contains = [“y”,”data”])
results = unittest.TextTestRunner().run(suite)

Plot containing a spatial raster image and spatial polygon vector data

from autograde.cases import PlotRasterSuite
axis = plt.gca()
suite = PlotRasterSuite(ax=axis, im_expected=image, polygons=polygons)
results = unittest.TextTestRunner().run(suite)

If you prefer to forgo the groupings into TestSuites, you can just use the assertions instead.

2D plot with x-axis label containing "x" and y-axis label containing "y" and "data"

from autograde.base import PlotTester
import pandas as pd
axis = plt.gca()
pt = PlotTester(axis)
data = pd.DataFrame(data={“x”:xvals, “y”:yvals})
pt.assert_xydata(data, “x”, “y”)
pt.assert_xlabel_contains([“x”])
pt.assert_ylabel_contains([“y”, “data”])

Plot containing a spatial raster image and spatial polygon vector data

from autograde.raster import RasterTester
From autograde.vector import VectorTester
axis = plt.gca()
rt = RasterTester(axis)
vt = VectorTester(axis)
rt.assert_image(im_expected=image)
vt.assert_polygons(polygons_expected=polygons)

Caveats: This repo likely misses edge cases of the many ways matplotlib plots can be created. Please feel free to submit bugs!

Contributors

• Kristen Curry

Contributing Breakers:

• Leah Wasser