genser 0.0.2

A set of functions for dataset dimension transformations

The GenSer package (i.e. Generalised Serialisation) contains the set of functions to perform the dimension transformation of the numerical dataset. To use the package you should have a dataset of non-negative integer values. Having n features in your dataset, you may transform it to m features and, after your work with data, return back to n features. The following functions available:

dim_step_down(data, powers): # data is a list of lists; powers is dictionary in form {n1:N1,n2:N2}, where n1, n2 are the indices of the features being unified and N1,N2 are the corresponding powers of the features.

dim_step_up(data, powers): # data is a list of lists; powers is dictionary in form {n1:N,n2:N2}, where n1, n2 are the indices of the features: n1 will be divided onto n1 and n2, and N,N2 are the corresponding powers of the features. The resulted data will have feature n1 of power N/N2 rounded above and n2 of power N2.

transform_to(data, m): # transforms data to the new dataset of dimension m; every step will be made with most appropriate features and the information about the step will be returned in dictionary with indexes as keys and powers as values

transform_out_down(data, story): # inverse function: performs backward transformation after dimensions growth by transform_to function; story is the dictionary given by transform_to

transform_out_up(data, story): # inverse function: performs backward transformation after dimensions decrease by transform_to function; story is the dictionary given by transform_to

Additional information available directly from the author by request on email [e-mail].