Python Programming for Biology: Bioinformatics and Beyond

Data discrimination

Now we will move on from clustering data to consider data discrimination, i.e. finding the

view  (projection)  of  the  data  that  gives  the  best  separation.  We  will  consider  two

examples.  The  first  is  principal  component  analysis  (PCA),  which  can  be  used  to  show

trends in a data set without first having to group it into categories, although clustering will

often be performed after PCA. The second example is linear discriminant analysis (LDA),

which  operates  on  two  categories  of  data  and  illustrates  how  rules  can  be  set  up  to

discriminate between the groups, e.g. to perform classification on unseen data.