IPython Notebook for HessianLLE. #3540
Conversation
|
I think we should put the content into the existing notebook |
| "from mpl_toolkits.mplot3d import Axes3D\n", | ||
| "\n", | ||
| "\n", | ||
| "mat = loadmat('usps.mat')\n", |
There was a problem hiding this comment.
you need to make sure this works when executed from its folde ron our servers. See the other notebooks for the relative path of the usps data
| { | ||
| "cells": [ | ||
| { | ||
| "cell_type": "code", |
There was a problem hiding this comment.
Rather than putting all content into a single cell, split it into multiple ones.
Then dont use source code comments to explain what you are doing, but the notebook markdown in separate cells
| "#print Xall_.shape\n", | ||
| "Yall = np.array(mat['label'].squeeze(), dtype=np.double)\n", | ||
| "Yall_ = np.resize(Yall, (1024, ))\n", | ||
| "def plot(data, embedded_data, colors='m'):\n", |
There was a problem hiding this comment.
please define functions in their own cells
Also give every function a proper name, and put a minimal docstring
| "Xall_ = np.resize(Xall, (256, 1024))\n", | ||
| "#The number of training examples have been reduced to make it feasible to run.\n", | ||
| "\n", | ||
| "#print Xall_\n", |
There was a problem hiding this comment.
please no commented out code in patches
| "\tplt.axis('tight'); plt.axis('off')\n", | ||
| "\tplt.show()\n", | ||
| "\n", | ||
| "from modshogun import StochasticProximityEmbedding, RealFeatures, HessianLocallyLinearEmbedding\n", |
There was a problem hiding this comment.
pls do
import modshogun as sg and do all imports in a single cell
| "\n", | ||
| "#print converter.get_reconstruction_shift()\n", | ||
| "\n", | ||
| "plot(data, embedded_data, colors)" |
There was a problem hiding this comment.
every plot needs a title and axis labels.
In addition, it should have some description of what we can see
|
ill close for now, feel free to reopen if you have updates |
This addresses the #2674
This notebook is a SHOGUN analog of the scikit tutorial given here. Note that I have included the tutorial only for one of the manifold learning methods i.e.
Hessian LLE. More would be included once this is reviewed better.The output of the notebook (along with the plot, of course) is here.
@karlnapf Review.