DistrictDataLabs / yellowbrick / 553
90%

DEFAULT BRANCH: develop
Ran
Jobs 2
Files 80
Run time 2min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
553

push

travis-ci-com

web-flow 
Color DispersionPlot markers based on target values (#592)

This merge revamps the DispersionPlot visualizer so that now it can now use colored markers to depict class association.  This includes the ability to specify labels and colormaps.  Additionally, these changes include new pytests as well as improved error handling.

Example here "yellowbrick/examples/rebeccabilbro/dispersion_plots.ipynb"

Commit Titles
---------------
* Establish coloring of markers based on target values
* Establish Dispersion Plot with color coded classes including a legend
* fix failed test caused by usage of generators as input
* Address PR comments

1. Fixed colormap docstring entry by changing wording to "Qualitative colormap for discrete targets"
2. To make it a bit easier to understand, we changed self.words to self.target_words and self.target_words_ to self.indexed_words
3. Changed name of text variable from 'text' to 'X' to make it a bit more similar to how input features are referenced in scikit-learn and in our other text visualizers (e.g. FreqDist and TSNE)
4. Because `self.offset` is only used in `_compute_dispersion` function it is prudent to change it from class attribute to regular variable `offset`
5. Modify docstring for the `fit` method to reflect the change in how the method expects input.  It has changed from a list of words to a list/generator of documents that in itself contains a list of words.
6. Because `self.points` and `self.points_target` are only used in `fit` function it is prudent to change it from class attribute to regular variable `points` and `points_target`
7. Create Test that raises if number of labels doesn't match the number of classes

* Make sure that a custom YellowbrickValueError is raised when a target word is not found in the corpus
* Change _check_missing_words loop from enumerate to range(len())
* adds an example notebook with my experimentations using the color-by-class version of Dispersion Plot
* fixed... (continued)

3150 of 3648 relevant lines covered (86.35%)

1.73 hits per line

Jobs
ID Job ID Ran Files Coverage
1 553.1 0
86.43
 Travis Job 553.1
2 553.2 0
86.35
 Travis Job 553.2
Source Files on build 553
Detailed source file information is not available for this build.