Hi,
as far as I know, pandas does not support yet the analysis of time series homogeneity and the detection of breakpoints.
Would be really great if such a feature could be implemented. What I am thinking about is similar to the following R-project, called bfast (http://bfast.r-forge.r-project.org/).
Would be great if some people would be interested in this.
Alex
Comment From: nipunbatra
Even i was interested in something similar, but like edge detection and similar stuff in time series.
Comment From: jreback
@bulli92 can you give a concrete example of what you are looking for?
Comment From: cpcloud
this looks like it's outside of pandas domain and could be a separate package that depends on pandas
Comment From: bulli92
@jreback : The link to the R-package contains already a useful picture which illustrates well the general concept. Practical applications could be rather broad as e.g. - detection of systematic changes in financial records - detection of abrupt changes in geophysical data records (e.g. temperature) - generation of maps of breakpoints from e.g. satellite data sets - ...
@cpcloud: I agree, might be also useful as an additional package which is using pandas.
Comment From: sinhrks
closing as out of scope.