rtchange

An implementation of real-time change-point detection in time series. (See this paper)

Dependencies

The required dependencies to use rtchange are

• Python >= 3.3
• NumPy

Installtion

$ pip install -U rtchange

Example

>>> from rtchange import Finder
>>> f = Finder()
>>> # `score` method returns the iterator of score of the samples.
>>> scores = f.score([0, 0, 0, 0])
<generator object Finder.score at 0x7faa143c4620>
>>> print(list(scores))
[0.0, 0.3938543257165724, 0.42281301105991476, 0.4718002708508469]
>>> print(list(f.score([1, 1])))
[1.0910027829221285, 1.2528868802573792]
>>> # You can also use `score_one` method for the one sample. 
>>> f.score_one(2)
1.4303444070633553

Samples

Jumping means

TOYOTA stock prices

Testing

$ virtualenv venv
$ source ./venv/bin/activate
$ pip install -U pip
$ pip install -e .[testing]
$ make test