Orange Forum • View topic - Support for hdf5 file format

Support for hdf5 file format

A place to ask questions about methods in Orange and how they are used and other general support.

Support for hdf5 file format

Postby dashesy » Sat Nov 24, 2012 0:03

I would like to read our file format (hdf5) to Orange, what is the best approach?

I was thinking of a new add-on that I can write (using PyTables but of course in a way that understands our layout) for our customers, but I am very new to Orange.

We have many channels of time-series of possibly vector data, and I would like to fully benefit from all the great tools in Orange, specially Orange-Bioinformatics, and visualization.

Any suggestion is appreciated
dashesy

Re: Support for hdf5 file format

Postby Ales » Tue Nov 27, 2012 17:24

dashesy wrote:I would like to read our file format (hdf5) to Orange, what is the best approach?

The best (and basically the only at the moment) approach is to basically copy the data from the hdf5 table into an Orange.data.Table. First you will have to create a new Orange.data.Domain to match the hdf5/pytable datatype descriptors. Take specific note of feature descriptor reuse (although at first you might just get your code working without this and think about this later).
Then simply iterate over the rows and create/append new Orange.data.Instance to the Orange data table.

This is fairly broad overview (and possibly not all that usefull). Please ask if you have any more questions.


Return to Questions & Support



cron