-
Open the file Wine.xls in Microsoft Excel.
This file contains 13 quantitative variables measuring the chemical attributes
of wine samples from 3 different wineries (the Type variable). The
objective is to assign a wine classification to each record..
In XLMiner�, click on Partition
data --> Standard Partition.
In the ensuing dialog box, select all variables in the Variables box and move
them to the "Variables in the partitioned data" box. Select
Specify percentages and enter 80% for the training set and 20% for
the validation set. Select OK.
In XLMiner�, select Classification, then Neural
Network (Multilayer feedforward) option. In the Neural network
dialog box, move the variable "Type" to the "Output variable"
box, and move the remaining variables to the "Input variables"
box.
A more detailed explanation of the above dialog box follows:
Variables: This box lists
all the variables present in the dataset. If the "First row contains
headers" box is checked, the header row above the data is used to
identify variable names.
Variables in input data:
Select one or more variables as independent variables from the Variables box
by clicking on the corresponding selection button. These variables constitute
the predictor variables.
Weight variable: Use this option
if you have data where there are multiple cases (objects) sharing the same
variable values, and the weight variable denotes the number of cases with
those values.
Output Variable: Select one variable as
the dependent variable from the Variables box by clicking on the corresponding
selection button. This is the variable being classified.
Click Next, and the following dialog
box appears. Here you specify the architecture for the neural network.
The second dialog box contains options to define
the network architecture. For this sample, accept the default values.
Details on these choices are explained below the dialog box.
Normalize
input data: Normalizing
the data (subtracting the mean and dividing by the standard deviation) is important to ensure that the distance measure accords equal
weight to each variable -- without normalization, the variable with the
largest scale will dominate the measure. Check this box.
Number
of hidden layers: Up to four hidden layers can be specified; see the
overview section for more detail on layers in a neural network (input, hidden
and output). Let us specify the number to be 1.
#
Nodes: Specify the number of nodes in each hidden layer. Selecting the number of hidden layers and the number of nodes is largely a
matter of trial and error.
#
Epochs: An epoch is one sweep through all the records in the
training set.
Step
size for gradient descent: This is the multiplying factor for the
error correction during backpropagation; it is roughly equivalent to the
learning rate for the neural network. A low value produces slow but
steady learning, a high value produces rapid but erratic learning.
Values for the step size typically range from 0.1 to 0.9.
Weight
change momentum: In each new round of error correction, some memory
of the prior correction is retained so that an outlier that crops up does not
spoil accumulated learning.
Error
tolerance: The error in a particular iteration is backpropagated only if
it is greater than the error tolerance. Typically error tolerance is a small
value in the range 0 to 1. The default value for error tolerance in
XLMiner�
is 0.01.
Weight
decay: To prevent
over-fitting of the network on the training data set a weight decay is used to
penalize the weight in each iteration, thus updating it by multiplying the
calculated weight by (1-decay).