First of all, I want to say "Hi" to the community as this is my first question in the forum. :)
I am new to WEKA and have the following problem with data preparation.
I want to separate set 1 and 2. The features for each item are calculated by taking two matrices A and B with integers and calculating the euclidian distance for each column. I have eight columns in total, making eight features. If the two matrices have identical values in the same column, the distance is 0. If they don't, the distance is greater than 0. My problem is now the case where both columns have only Zeros as entries and the calculated distance is also 0. This case happens quite often. So if i put in 0 for the feature in both cases, the Zeros from the latter case (where both columns had only Zeros) totally cover the very few cases where the 0 in the feature comes from equal values in the matrices. It is important for my application to separate between these two cases.
What I now did is choosing the value as missing for the case the 0 comes from two columns with only 0s. Would it be better to choose another value, like -1 or -5? I would like to use trees and logistic regression for classification. Or does it not make any difference at all?
Thank you studentdd
0 answers
No answers yet.
Log in to answer this question.
-
Dealing with 2 different layers of counts within RNA assay in seurat
written by Varun Gupta
-
How to exclude features that intersect with other features in the same file?
written by George •
-
Hierarchical clustering of non-normal data
written by Jules •
-
Meaning of header in TCGA metadata
written by foehn •
-
Difference between heatmap using Euclidian distance and Poisson distance
written by tlorin
-
Mantel tests results
written by eng_samar_zayed •
-
Machine learning based classification techniques
written by tigeradab
-
PCA for three dimensional features (or for 3 matrices)
written by pbio
-
Pvclust:Error In Clustering
written by Diana
-
Machine Learning To Find The More Informative Features To Predict And Outcome
written by Lorena Pantano
Please edit this question to indicate relevance to a bioinformatics research problem.
Weka is obviously relevant to bioinformatics research. Anyone in the community, who understands something about it, could add a nuance or example that would satisfy Neilfws's request (i.e. i.e. given the importance and relevance of machine learning, it doesn't have to necessarily fall on the responsibility of the poster). I really wish we could get away from rote "define relevance" responses when it comes to things with such obvious utility in bioinformatics.