can anyone tell me whether the real attribute is the continuous attribute or not?
if yes, any suggestion about discritizing these attributes
Yes, real numbers are continuous numbers.
search for discretization features which is the process of converting real features to nominal features.
this is a nice review about Discretization Techniques: http://www.math.upatras.gr/~esdlab/oldEsdlab/en/members/kotsiantis/discretization%20survey%20kotsiantis.pdf
"Real" is probably a concept taken from various computer languages,
which describes a continous variable (in your case an attribute).
A paper entitled
"REAL: Real Attribute Learning for Strategic Market Operation"
See at:
http://www.ime.usp.br/~jstern/papers/papersJS/REAL.pdf
deals also with Discretization Procedure (jn Chapter 5)
Yes, 'Real' attribute is a continuous attribute, as opposed to a discrete attribute-- which may be binary (0 or 1), nominal (e.g. red, blue, green), or ordinal (e.g. high, medium, low).
There are various ways to discretize a real attribute into suitable number of categories, but I think that is best done depending on some Class attribute (in case the problem is a classification problem), or a target variable (in a general machine learning problem). This is generally called "Supervised Learning."
The task of discretizing an attribute on an "Unsupervised" basis is a lot more difficult.
A very popular paper on this is by Liu, Hussain, Tan and Dash.
http://www.cs.sfu.ca/cc/459/han/papers/hussain99.pdf
Article Discretization: An Enabling Technique
Thanks a lot for your answer sir Partha. could you support me with paper discussing the fuzzy discretizatio , i think it's appropriate for the DNA microarray data which i'm working on
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining