Comments (4)
There is a special undocumented option I added awhile back to help shift the cover bins a bit, so that points near the boundaries were accounted for properly. See scale_limits=True
in the following line (executed cell 11 in the notebook):
cover = optimize_cover(
X, r=15, g=0.67,
scale_r=not True,
scale_limits=True
)
The corresponding code is defined in dyneusr/mapper/utils.py, see e.g.,
# Define optimized limits
limits = None
if scale_limits is True:
offset = p_overlap / float(n_cubes)
limits = [[-offset, 1+offset] for _ in range(ndim)]
n_cubes += 2 #* ndim
Not 100% sure why this doesn't work with kmapper
anymore, but I will take a closer look and see if there's an obvious way to fix this. Otherwise, the easiest thing to do may be to just turn it off in the tutorial, and possibly deprecate the option altogether.
Edit: To clarify, when I use scale_limits=False
, I get a graph with 2007 edges and 211 nodes, for the first subject.
from dyneusr.
For more context, when I use scale_limits=True
, I get the following:
>>> cover.limits
array([[-0.04466667, 1.04466667],
[-0.04466667, 1.04466667]])
>>> cube_centers = cover.fit(lens)
>>> cube_centers
[array([-0.01262745]),
array([0.05145098]),
array([0.11552941]),
array([0.17960784]),
array([0.24368627]),
array([0.30776471]),
array([0.37184314]),
array([0.43592157]),
array([0.5]),
array([0.56407843]),
array([0.62815686]),
array([0.69223529]),
array([0.75631373]),
array([0.82039216]),
array([0.88447059]),
array([0.94854902]),
array([1.01262745])]
>>> hyper_cubes = cover.transform(lens, cube_centers)
>>> [_.shape for _ in hyper_cubes]
[(3, 2),
(2, 2),
(1, 2),
(1, 2),
(1, 2),
(2, 2),
(2, 2),
(3, 2),
(3, 2),
(4, 2),
(4, 2),
(4, 2),
(3, 2),
(2, 2),
(1, 2)]
and when I use scale_limits=False
:
>>> cover.limits
None
>>> cube_centers = cover.fit(lens)
>>> cube_centers
[array([-17.97072105]),
array([-15.59127478]),
array([-13.2118285]),
array([-10.83238223]),
array([-8.45293595]),
array([-6.07348968]),
array([-3.6940434]),
array([-1.31459713]),
array([1.06484915]),
array([3.44429542]),
array([5.82374169]),
array([8.20318797]),
array([10.58263424]),
array([12.96208052]),
array([15.34152679])]
>>> hyper_cubes = cover.transform(lens, cube_centers)
>>> [_.shape for _ in hyper_cubes]
[(26, 2),
(40, 2),
(48, 2),
(47, 2),
(47, 2),
(46, 2),
(50, 2),
(58, 2),
(60, 2),
(56, 2),
(49, 2),
(47, 2),
(51, 2),
(50, 2),
(32, 2)]
So it seems to me like it has to do with the scaling of these limits or the lens itself.
from dyneusr.
@kristiandroste looping you in here
from dyneusr.
Thanks @calebgeniesse!
The notebook runs start to finish without catching any errors when I use scale_limits=False
, but it seems something is still not right with my graphs. I document some discrepancies below and will try debugging further
The shapes of my hyper_cubes
and my cube_centers
differ from those you posted above:
Running
cube_centers = cover.fit(lens)
print(cube_centers)
hyper_cubes = cover.transform(lens, cube_centers)
shapes = [arr.shape for arr in hyper_cubes]
print(shapes)
at the end of cell 12 yields
# cube_centers
[array([-15.88660342]),
array([-13.70202765]),
array([-11.51745188]),
array([-9.33287611]),
array([-7.14830033]),
array([-4.96372456]),
array([-2.77914879]),
array([-0.59457302]),
array([1.59000275]),
array([3.77457852]),
array([5.95915429]),
array([8.14373006]),
array([10.32830583]),
array([12.51288161]),
array([14.69745738])]
# shapes of hyper_cubes
[(26, 2),
(41, 2),
(47, 2),
(50, 2),
(54, 2),
(57, 2),
(60, 2),
(61, 2),
(60, 2),
(55, 2),
(45, 2),
(43, 2),
(43, 2),
(43, 2),
(27, 2)]
Also, the output from cell 11 shows that I'm getting the same shape of X
for all subjects:
X has shape: (242, 577)
Comparing this to the shapes listed in the screenshot of the output in the DyNeuSR Notebooks repo, it appears that this shape is correct for subject 1, but the other five subjects should have different shapes for X
.
Visualization of mapper (cell 14), heatmaps (cell 16), and TCM visualizations in step 6 all differ from the output in the DyNeuSR Notebooks repo
from dyneusr.
Related Issues (11)
- docker app HOT 6
- test failures with TestDyNeuGraph.test_visualize and TestDyNeuGraph.test_visualize_show HOT 12
- Problem importing dyneusr; matplotlib TkAgg backend HOT 8
- Issues converting to DyNeuGraph HOT 2
- Saving a shape graph in the HTML file as SVG automatically
- deprecated import from sklearn HOT 2
- Use versioned requirements when deploying dyneusr to pip repo
- visualize_mapper_stages error in examples HOT 4
- DyNeuGraph error due to networkx changes HOT 3
- problem in ds.tools.networkx_utils.visualize_mapper_stages function
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from dyneusr.