toady

Easily visualize high-dimensional data in 2d space.

Installation

pip install toady

Basic Usage

Below is a very simple example using the Iris dataset.

import pandas as pd
from toady import toady

data = pd.read_csv('iris.csv')

features = ['SepalLengthCm', 'PetalLengthCm', 'PetalWidthCm']
X = data[features]
y = data['Species']

toady(X, y)

Example with point labels

Below we map 7 features of the world's top universities onto 2d space and color the points based on university score. We also add a label for each point:

data = pd.read_csv('cwurData.csv')

features = [
    'quality_of_education',
    'alumni_employment',
    'quality_of_faculty',
    'publications',
    'influence',
    'citations',
    'patents',
]

X = data[features]
y = data['score']
labels = list(data['institution'].values + ' (' + data['year'].apply(str).values + ')')

toady(X, y, labels)

In this embedding, the very top schools in the world (e.g. Harvard, Princeton, etc.) are actually near each other in our embedding (hover on points to reveal labels):

Customizability

Other parameters such as the embedding model used, the scaling model used, etc. may be adjusted.

Refer to the parameter list below for more information on adjusting these parameters.

Parameters

• X : { pd.DataFrame }

  DataFrame containing input data / predictors / features. (Can contain categorical, missing data, etc)

• y : { pd.Series }

  Series containing the target variable. (Can be categorical)

• point_labels : { list }

  List containing the label of each scatter plot point.

• impute_model : { type }, default 'sklearn.preprocessing.Imputer'

  Model used for imputing missing values in the data.

• impute_params : { dict }, default empty dict

  Params fed to impute_model.

• scale_model : { type }, default 'sklearn.preprocessing.RobustScaler'

  Model used for scaling the data.

• scale_params : { dict }, default empty dict

  Params fed to scale_params.

• embed_model : { type }, default 'sklearn.manifold.Isomap'

  Model used for embedding the data onto 2d space.

• embed_params : { dict }, default empty dict

  Params fed to embed_model.

• scatter_params : { dict }, default empty dict

  Params fed to the scatter plot.

• css : { CSS string }, default seen in README

  CSS string for tooltips.

• verbose : { bool }, default False

  Whether or not informative messages are shown at each step of the toady process.