This project is under BSD license.


With pip3 from terminal: $ pip3 install ruptures.

Or download the source codes from latest release and run the following lines from inside the folder $ python3 install or $ python3 develop.

User guide

This section explains how to use implemented algorithms. ruptures has an object-oriented modelling approach: change point detection algorithms are broken down into two conceptual objects that inherits from base classes: BaseEstimator and BaseCost.

Initializing a new estimator

Each change point detection algorithm inherits from the base class ruptures.base.BaseEstimator. When a class that inherits from the base estimator is created, the .__init__() method initializes an estimator with the following arguments:

  • 'model': “l1”, “l2”, “normal”, “rbf”, “linear”, “ar”. Cost function to use to compute the approximation error.
  • 'cost': a custom cost function to the detection algorithm. Should be a BaseCost instance.
  • 'jump': reduce the set of possible change point indexes; predicted change points can only be a multiple of 'jump'.
  • 'min_size': minimum number of samples between two change points.

Making a prediction

The main methods are .fit(), .predict(), .fit_predict():

  • .fit(): generally takes a signal as input and fit the algorithm on the data
  • .predict(): performs the change point detection. This method returns a list of indexes corresponding to the end of each regimes. By design, the last element of this list is the number of samples.
  • .fit_predict(): helper method which calls .fit() and .predict() successively.

Creating a new cost function

In order to define custom cost functions, simply create a class that inherits from ruptures.base.BaseCost and implement the methods .fit(signal) and .error(start, end):

  • The method .fit(signal) takes a signal as input and sets parameters. It returns 'self'.
  • The method .error(start, end) takes two indexes 'start' and 'end' and returns the cost on the segment start:end.

An example can be found in Custom cost class.