How to Upload Data File
Here is a picture of the page that enables data upload and start of the
induction process. With START button you actually start data upload
and automatically after it data mining process. You will have to wait some
time before results are presented on your screen. Computation time depends on the
problem complexity and the number of other users on the server. Typical
computation time is 10 seconds. But before you press button START you must
enter the name of your data file and select some options.
A select the name of the data file on your computer. Instead of typing
the name and the path of the file, you can press 'Browse...' button and search for
the data file. The name of your data file is completely your choice.
B specify the
type of the delimiter you have used in the data file. Selecting a wrong
delimiter will prevent the server to correctly interpret the
data file and, typically, strange errors about inappropriate number of input attributes
will be returned to you.
C select 1-3 models
that will be generated by the server. Please note that
the requested number of models is presented only if the system actually can
find as much good models. In cases when number of submitted examples is small or when the
first model is significantly better than all other models, it can happen that you receive
one or two models even you requested three.
D select the generalization parameter
which enables to user to influence the induction process. Larger generalization
parameter values will tend to construct models of larger subgroups
for the positive target class often with the result that such models may incorrectly
include also some cases of the negative class. Small parameter values will result in
specific models which are correct for small subgroups of the positive class.
E additionally select if you want
noise detection in the preprocessing.
This selection will not influence the type of models that will be induced but it may give
you some additional information about your data set. Do not use this option if not necessary
because it can significantly increase the computation time for your task.
Go to data upload page.
© 2001 LIS - Rudjer Boskovic Institute
Last modified: March 23 2019 15:30:09.