Ticket #17 (closed wish: wontfix)

Opened 7 years ago

Last modified 6 years ago

Numpy: pretvorba meta atributov iz ExampleTable v numpy in obratno

Reported by: janez Owned by: janez
Milestone: Future Component: library
Severity: minor Keywords:
Cc: Blocking:
Blocked By:

Description

Pretvorba ExampleTable->numpy ne uposteva vrednosti v meta-atributih, pretvorba numpy->ExampleTable ne zna pisati v meta-atribute.

Obvoz: pretvori meta v navadne atribute:

et1 = orange.ExampleTable(orange.Domain(orange.FloatVariable("p-vals")), numpy.ones((10,1))) d2 = orange.Domain([]) d2.addmeta(orange.newmetaid(), et1.domain.variables[0]) et2 = orange.ExampleTable(d2, et1) et2[0]

[], {"p-vals":1.000}

To je nerodno. Problem so tudi sparse matrike.

Change History

comment:1 Changed 7 years ago by janez

  • Component set to core
  • Milestone changed from 1.0 to future

comment:2 Changed 6 years ago by janez

  • Status changed from new to closed
  • Resolution set to wontfix

Za "sparse meta atribute" bi bil primeren kvecjemu sparse matrix. Ta je v scipy, ki pa vsebuje pretezno stvari, ki nas ne zanimajo, velik pa je 11 MB, vec kot ves Orange skupaj.

Za ne-sparse zgornja finta zadovoljivo deluje.

Morda bomo nekoc v prihodnosti migrirali v numpy (tako, da bo ExampleTable izpeljan iz ndarraya ali kaj podobnega), zato trenutno nima smisla kaj veliko delati na tem.

Note: See TracTickets for help on using tickets.