dalys 4.1.0

Dalys

Dalys - library for visualising data analysis algorithms

Dalys - library that allows to visualize many algorithms that need for statistical data processing and machine learning

Getting Started

Dalys is a wrapper over the library scikit-learn and you can use scikit-learn datasets for visualize desired algorithm.

Example:

from sklearn.datasets import make_circles
from dalys.kernel_pca import KPCATool


X, y = make_circles(n_samples=1000, random_state=123, noise=0.1, factor=0.2)

style = [('red', '.'), ('blue', '.')]
label_names = ['red dotes', 'blue dotes']

kpca = KPCATool(X, y, style=style, labels_unique_name=label_names, n_components=3,
                kernel='rbf', gamma=2, scaled=True)
kpca.projections_plot(grid=22)

There is a number of parameters that can be set for the KPCATool class; the major ones are as follows:

• style - set marker styles and colors for data representation, example: style = [('red', '.'), ('blue', '.')] or style = [('red', 'd'), ('blue', 'd')]. If you need random colors and markers - leave this argument with default parameter (default = None).

• labels_unique_name - set list with class names, such as labels_unique_name = ['red dotes', 'blue dotes']. If labels_unique_name is None, this means that labels will be numbered in ascending order (default = None).

• preprocessing - for use data preprocessing, set this parameter to one of the following options (default = 'std'):

  • std - standardize features
  • norm_l1 - L1 normalization
  • norm_l2 - L2 normalization
  • minmax - MinMax scaler
  • maxabs - MaxAbs scaler

• scaled - this flag need to control preprocessing of your data, if your data is allready processed, you must set "True".

Installing

pip install dalys

Authors

• Timothy Tkachenko - Machine Learning Researcher