Confused about how to graph my high dimensional dataset with Kmeans

So I have a dataset that is very high dimensional, and I am very confused about to convert it into a form that can be used to plot with Kmeans. Here is an example of what my dataset looks like:

Blk Students %       White Students %     Total Endowment    Full time Students     Tuition Revenue

35                    60               $$4,000,000 50000$$4,999,999
50                    50               $$5,888,888 67899$$200,000,000
.                     .                    .                     .
.                     .                    .                     .

.

I have normalized my variables, converted everything to integers, etc... But I am confused about how I am supposed to take this high dimensional dataset and plot it on a 2D scatterplot for kmeans? Thanks in advance.

• Is this the entirety of your columns? Oct 20 '18 at 19:18

Your k-means should be applied in your high dimensional space. It does not need to be applied in 2D and will give you poorer results if you do this. Once you obtain the cluster label for each instance then you can plot it in 2D. However, we live in a 3D world thus we can only visualize 3D, 2D and 1D spatial dimensions. This means you can at most plot 3 variables in a spatial context, then you can maybe use the color of your points as a fourth dimension. If you really want to stretch it you can use the size of your points for a 5th dimension. But these plots will quickly get very convoluted.

You can use dimensionality reduction techniques to project your high dimensional data onto 2 dimensions. What you are looking to do is perform some projection or feature compression (both of those terms mean the same thing in this context) onto a 2D plane while maintaining relative similarity. Many of these techniques exist each optimizing a different aspect of relative "closeness".

The rest of this answer is taken from here.

The following code will show you 4 different algorithms which exist which can be used to plot high dimensional data in 2D. Although these algorithms are quite powerful you must remember that through any sort of projection a loss of information will result. Thus you will likely have to tune the parameters of these algorithms in order to best suit it for your data. In essence a good projection maintains relative distances between the in-groups and the out-groups.

The Boston dataset has 13 features and a continuous label $$Y$$ representing a housing price. We have 339 instances.

from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
%matplotlib inline
from sklearn.manifold import TSNE, SpectralEmbedding, Isomap, MDS

X = boston.data
Y = boston.target

X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.33, shuffle= True)

# Embed the features into 2 features using TSNE# Embed
X_embedded_iso  = Isomap(n_components=2).fit_transform(X)
X_embedded_mds  = MDS(n_components=2, max_iter=100, n_init=1).fit_transform(X)
X_embedded_tsne = TSNE(n_components=2).fit_transform(X)
X_embedded_spec = SpectralEmbedding(n_components=2).fit_transform(X)

print('Description of the dataset: \n')

print('Input shape : ', X_train.shape)
print('Target shape: ', y_train.shape)

plt.plot(Y)
plt.title('Distribution of the prices of the homes in the Boston area')
plt.xlabel('Instance')
plt.ylabel('Price')
plt.show()

print('Embed the features into 2 features using Spectral Embedding: ', X_embedded_spec.shape)
print('Embed the features into 2 features using TSNE: ', X_embedded_tsne.shape)

fig = plt.figure(figsize=(12,5),facecolor='w')
plt.subplot(1, 2, 1)
plt.scatter(X_embedded_iso[:,0], X_embedded_iso[:,1], c = Y, cmap = 'hot')
plt.title('2D embedding using Isomap \n The color of the points is the price')
plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.colorbar()
plt.tight_layout()

plt.subplot(1, 2, 2)
plt.scatter(X_embedded_mds[:,0], X_embedded_mds[:,1], c = Y, cmap = 'hot')
plt.title('2D embedding using MDS \n The color of the points is the price')
plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.colorbar()
plt.show()
plt.tight_layout()

fig = plt.figure(figsize=(12,5),facecolor='w')
plt.subplot(1, 2, 1)
plt.scatter(X_embedded_spec[:,0], X_embedded_spec[:,1], c = Y, cmap = 'hot')
plt.title('2D embedding using Spectral Embedding \n The color of the points is the price')
plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.colorbar()
plt.tight_layout()

plt.subplot(1, 2, 2)
plt.scatter(X_embedded_tsne[:,0], X_embedded_tsne[:,1], c = Y, cmap = 'hot')
plt.title('2D embedding using TSNE \n The color of the points is the price')
plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.colorbar()
plt.show()
plt.tight_layout()

The target $$Y$$ looks like: The projected data using the 4 techniques is shown below. The color of the points represents the housing price.  You can see that these 4 algorithms resulted in vastly different plots, but they all seemed to maintain the similarity between the targets. There are more options than these 4 algorithms of course. Another useful term for these techniques is called manifolds, embeddings, etc.

Check out the sklearn page: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.manifold.

There are a number of ways to plot high dimensional data. Matplotlib supports 3D plotting, so I assume you won't be having a problem plotting the 3 dimensions on x, y, and z-axis of the graph. However, for higher dimensions, you can use colors and symbols. For instance, if you plot normalized full-time students along x-axis, total endowment as along y-axis, and tuition revenue as z-axis on a 3D plot, then you can plot, for instance, a colored triangle for the %ge of your first column, and the color can be a heat map value, where blue shows low percentage, red shows a high percentage. Similarly, you can use a color-map square, or a circle to show plots of your second column.

This is, no doubt, one idea to do it. But other techniques can be, that you perform some dimensionality reduction and visualize meaningful information, or you can divide a high-dimensional data into several occurrences of a lower dimensional data, and plot all of them as subplots of a single large plot.

I've had good success using Linear Discriminant Analysis (LDA) in two dimensions to plot multi-dimensional clustering results. LDA finds the projection that maximizes the separation between the means of the clusters. For example, here is a 2D projection of an 8-dimensional k-means clustering problem You can use LDA in any number of dimensions, and so if 3D renderings are desired LDA can be applied. I personally have a hard time making heads or tails of 3D scatter plots, though.