ANN Project For this project you will develop an ANN that mimics the diabetes di

ANN Project
For this project you will develop an ANN that mimics the diabetes diagnosis decision tree and expert system projects.
NOTE: As stated in the assignment, you may use a different neural network program, so long as it can produce a picture of your trained neural network and a picture of your training error (Root Mean Square Error of training). The rest of this explanation is for the JustNN program.
First, download and and install the JustNN free ANN shell tool from:
Remember where it gets installed, typically as a folder in your Program Files (x86) directory.
Now it is time to create your training (and validation) data files.
Use the Excel data file that you previously used for the Decision Tree and Expert Ssytem projects, DiagnosisSampleData.xlsx. 
Decide which input variables you want to use.
Divide the data into two sets, one for training and one for validating your ANN.
Move the Diagnosis column or columns to the first column (A), as this is the position needed for JustNN.
Save each spreadsheet as a comma separated text file.
Verify that the file saved correctly and that the first row is the names of each variable
Now move your comma separated text file into the JustNN directory
Time to start building your ANN
Find the JustNN application in the JustNN directory and start the program (double-click)
Click to say you want to use the JustNN program (and not either of the two pay-for programs)
After the JustNN program starts and the JustNN window appears, click on the file menu and select New project (unless you are returning, then you can open an existing project)
Click on the Import File Icon (an open, blue folder) and select your text file.
On the import screen that will open up, make sure the comma delimiter box is checked, the use numbers for row names, and the training box are also checked.
On the next screen click Set Names button, but this requires that the first linie of your training data file have the variable names, such as: Diagnosis,vomiting,diaphoretic, etc…
Next you will see what looks like a spreadsheet and you will b asked to identify the types of data being imported.  For the very first column, make sure that the output option is selected (input will be used for all remaining columns) as well as the desired type (text).
Continue with the remainder being Input variables and select the correct data type for each.  When done you will see a table that looks like what you saved as the text file from Excel.  Be sure to save this file in JustNN as a .tvq network file.
Click on the Grow Network icon to start building your ANN.  It will ask you how many hidden layers you want and what the maximum number of hidden nodes per layer can be.  Make sure that the grow layer number 1 is checked.  You may optionally check the 2 and 3 for second and third hidden layer boxes.  Next enter the number of nodes per layer maximum.  For the first hidden layer this should be twice the number of variables you are using. If you use a second hidden layer this should be the same number as the number of variables you are using (or half the first layer value).  Once these values are entered, click on Okay and then click Yes in the next pop up window.
Finally you will be asked for some control information for the ANN. You can leave these alone, but the boxes on the right may be useful to adjust to make sure your ANN stops.  An alternate way is to click on the Stop icon, which looks like an arrow pointing right to a bunch of lines. You can select one of the options in the lower right, Fixed period stops, to force the ANN to stop learning sooner.
Click the View Network icon (three to the left of the Grow Network icon) to see your ANN!  After training has stopped (or you have stopped training) do an ALT-Print Screen to capture what your network looks like. Save this to a Word document, with an explanation of why you chose the input variables you did.
Now see what the learning rate for your ANN looked like by clicking on the View Learning icon (looks like several squiggly lines in a graph).  Capture this screen also and save to your Word document.  Describe the learning that occurred, what could explain the shape of the average error rate for your ANN?
If you want to test your ANN, you can add queries to your data grid by either clicking on the +Q (Add Query) icon or select Add Query from the Query menu item.  This will add additional rows to your grid marked as queries and can be used to test the grid with the validation examples you saved out.
What you must turn in for this project is the document containing the pictures of your finished network and the corresponding learning rate graph and the screen showing the variable importance, along with your explanations for why you chose the variables you did (make sure you have a well thought out reason) and a comparison of the See5 decision tree versus the JustNN neural network.    Please feel free to try alternate input variable selections and also different hidden layer architectures (# of layers and/or # of nodes).  Attach an image of any additional ANNs you create using JustNN, with variable importance.
May i use the expert Adriel009