Handle missing values in data sets

GoogleCodeExporter commented 9 years ago

From loading through the SOM algorithm, IFC Soft should allow data points to 
have missing values. Currently, when reading a file, if it finds a missing 
value, it throws out the data point.

In the SOM algorithm, if a value is missing, the data point can still be placed 
based on the dimensions it has, and it should then only modify those dimensions 
which it has.

For WRCs, perhaps the wedge could be given a size of 1/2, but somehow be marked 
as N/A.

Original issue reported on code.google.com by kyle.tha...@gmail.com on 13 Apr 2011 at 8:13

GoogleCodeExporter commented 9 years ago

I started a thread to discuss how to solve this and issue 10:
http://groups.google.com/group/ifcsoft/browse_thread/thread/9b596a640401520b

Original comment by kyle.tha...@gmail.com on 24 May 2011 at 9:10

GoogleCodeExporter commented 9 years ago

Progress on this has been started on this clone: 
http://code.google.com/r/kylethayer-breakable/
I'm thinking it would be good to give an option to use 
1) All data points regardless of what they are missing
2) Only data points with at least half of the used dimensions
3) Only use data points that have all the used dimensions

Original comment by kyle.tha...@gmail.com on 21 Jul 2011 at 7:06

kylethayer / ifcsoft

Handle missing values in data sets #9