usualsuspects package

usualsuspects.tsne module

class usualsuspects.tsne.Quick2DTSNE(X, y, perplexity, initialisation='pca', n_jobs=-1)[source]

A quick tool for creating 2D T-SNE plots with input X, and optional true classifications y for coloring.

From Scikit Docs: t-Distributed Stochastic Neighbour Embedding is a tool for visualizing high-dimensional data. It converts similarities between data points to joint probabilities and tries to minimize the KL divergence between the joint probabilities of the low dimensional embedding and the high-dimensional data. T-SNE has a cost function that is not convex. i.e. with different initializations we are bound to have different results.

This tool is a shortcut to producing the T-SNE plot one finds so often in papers.

Parameters:
  • X (array-like) – asdfasdf
  • y (array-like) – asdfasdf
  • perplexity (float) – The perplexity is a hyperparameter that is related to the number of nearest neighbours to consider. Typically larger datasets need a higher perplexity to look good.
  • initialization (str (default="pca")) – Possible values = [‘pca’, ‘random’] describes the initialization.
  • n_jobs (int (default=-1)) – The number of cores to use. Default is -1 which uses all available system cores
Returns:
Return type:

Quick2DTSNE
plot_embedding(title=None, save_to='tsne_plot.png')[source]

Creates a matplotlib pyplot and saves it as a png image.

Parameters:
  • title (str) – Title for the plot
  • save_to (str) – path and filename where the image will be saved. Extension is contextual but expects .png or .jpg
Returns:

Saves a PNG image of the plot to the path in save_to
Return type:

None

usualsuspects.pca module

class usualsuspects.pca.Quick2DPCA(X, y)[source]

A quick tool for crating 2D PCA plots with input X, and optional true classification y for colouring

From SciKit Docs: Linear dimensionality reduction using SVD of the data to project it to a lower dimensional space. The input data is centered but not scaled for each feature before applying the SVD

This tool is a shortcut to producing PCA plots

plot_embedding(title=None, save_to='pca_plot.png')[source]

Creates and saves a matplotlib pyplot of the first 2 eigenvectors of the PCA operation on X

Parameters:
  • title (str) – Optional title for the plot
  • save_to (str) – path and file_name where the image will be saved. Extension is contextual but expected behaviour is to use .png
Returns:

Saves a PNG image of the plot to the path in save_to
Return type:

None