What is the Horizontal axis represent in R value figure in the ANN fitting?

I am using the fitting app in Matlab ANN (nnstart), but I am confused about the R value figure generated by the tool after traing. What is the Horizontal axis represent in R value figure in the ANN fitting? For example, the attached figure is R value of a ''body fat'' training example. Why the x-axis range is 0-40 or 0-50? It is not the sample number? How does the x-axis (Target) is obtained? Could anyone help me? Thanks.

Answer 1

[Aditya Patil]

The plot shown is created using plotregression function. It shows relation between the targets on the y axis, and the predicted output on the y axis. the axis limits are based on the limits of these two values. See the following code for example.

targets = randi(100, [1 100]); % expected output
outputs = targets + randi(10, [1 100]); % predicted output
plotregression(targets,outputs)

[Sunny]

Thank you Aditya.

[Sunny]

Aditya, I have a follow up question. Could you help me with that? If the output is a multi-dimentional vector, how will the Target be calculated? For example, the build dataset in example dataset of the fitting app in Matlab ANN (nnstart). The attached figure is the obtained R-value of this example. How dose it convert the 3-dimentional vector to a 1-dimentional target? What type of calculation does it apply? Do you know where I can find some information on this issue? Thank you in advance.

Here are the information of the build dataset in matlab help doc.

building_dataset Building energy dataset.

Function fitting is the process of training a neural network on a

set of inputs in order to produce an associated set of target outputs.

Once the neural network has fit the data, it forms a generalization of

the input-output relationship and can be used to generate outputs for

inputs it was not trained on.

This dataset can be used to train a neural network to estimate the

energy use of a building from time and weather conditions.

LOAD building_dataset.MAT loads these two variables:

buildingInputs - a 14x4208 matrix defining fourteen attributes for 4208

different houses.

1-10. Coded day of week, time of day

11. Temperature

12. Humidity

13. Solar strength

14. Wind

buildingTargets - a 3x4208 matrix of energy usage, to be estimated

from the inputs.

[X,T] = building_dataset loads the inputs and targets into

variables of your own choosing.

For an intro to fitting with the Neural Fitting app

click "Load Example Data Set" in the second panel and pick this dataset.

Here is how to design a fitting neural network with 10 hidden neurons

with this data at the command line. See fitnet for more details.

[x,t] = building_dataset;

net = fitnet(10);

net = train(net,x,t);

view(net)

y = net(x);

[Aditya Patil]

With multiple outputs, you should plot independent plot for each output-target pair. A combined plot simply takes all the values, which won't be meaningful. For example, if one output is age and another output is weight, it doesn't make sense to plot age and weight on the same axis, and fit a regression to it.

[Sunny]

That makes sense. Thank you Aditya.