Updated examples to plot PCA with Roassal 3 #321
Conversation
|
Polymath has no dependency on Roassal and I don't think it is a good idea to depend on a visualization framework for a math framework. Maybe @olekscode had an opinion? |
|
I agree with Cyril, we should not put a dependency to Roassal |
|
Could we have a Baseline group for loading visualizations examples as extension methods and so on? |
|
In my opinion, this should be in the data-inspector repo since we have there other visualizations as well |
|
Can we move all PolyMath visualisations in the PolyMath org? |
|
I'm not against creating a new repo inside PolyMath for visualizations. But, we should be careful to not have duplicated visualizations with data-inspector. |
|
I'm also not against it. For me data-inspectors is really about visualizations to help take decisions about your data. If we do on for PolyMath, IMA it should be things helping to visualize Polymath related things. Not visualizations to help take decisions over datas |
* master: refactor: Rename Variable where we make the roles of the objects clearer in the tests. test: clarified the name of a test. fix: corrected a comment. test: clarified naming using domain language (repeated roots). fix: corrected the comment. test: added some interesting and missing tests. test: changed the example and added a comment to make it clear what the coefficients correspond to. test: added a simple test for x^2 + 2x + 1, whose roots are x = -1 and -1 refactor: Rename Variable - it looks like it is a root. stylg: removed redundant comment. refactor: Rename Method, and now we have the interface that a client (in this case the test) needs. refactor: Inline Method as it has no clients. style: corrected a Smalltalk code critique. refactor: clarified the name of the method. refactor: Inline Method, as it has no clients. refactor: Move Method to a more sensible place. refactor: made the method a class method. Next we can inline the instance method. refactor: Extract Method where a method instantiates a root finding algorithm with the desired precision. refactor: migrated the client to no longer use roots message. refactor: using Parallel Change, introduced a new message that returns roots in order.
Update baseline dependencies. Also reduce baseline long method. Update Roassal deprecated class names.
|
I repackaged it in a new package |
|
You have two approaches
So the second one cost less in terms of management. And I think that in anycase this is cool to have visualizations :) |
|
Yes, this PR fixes the dependency problem. If you load it normally PM packages are loaded without any visualization dependency: Metacello new
repository: 'github://PolyMathOrg/PolyMath:plot_pca';
baseline: 'PolyMath';
loadSo there is no Roassal dependency (except a Roassal reference from another package, "Math-TSNE", which I will address in another PR). If you load the branch of this PR, there will be a new package Metacello new
repository: 'github://PolyMathOrg/PolyMath:plot_pca';
baseline: 'PolyMath';
load: #('visualizations')I tested under Pharo 11 and Pharo 12 and it works as expected. 
It is already safe to be integrated |
|
It looks good for me |
This needs PolyMathOrg/DataFrame#237 to be merged.
