Artificial Neural Network prediction

I am Sumit Kumar Ghosh. I have a question regarding prediction by Artificial Neural Network.

If I want to predict that what will be the value of ph in 2020. What I have to do in Artificial Neural Network?

Data:

x= Ph; which range is 0 to 14.

Lets,

x=[ 6,5,8] y=[ 2014,2016,2017]

I have to find what will be the value of x in the year of 2020 by the Artificial Neural Network.

I can't find any perfect solution. Will you please help me?

Sincerely, Sumit Kumar Ghosh

[Grzegorz Knor]

You have only three data points. It is not possible to use artificial neural network properly in this case. You might need to use a larger data set. To get started with neural networks in MATLAB please use nnstart.

[Greg Heath]

You can fit a parabola to three points. Typically, neural networks are used to obtain models based on tens, hundreds and thousands of points.

Hope this helps.

Greg

[Sumit Kumar Ghosh]

I have this data set where there are 20 parameters for 10 locations of 2017. how i will predict 20 parameters for these 10 locations for 2018,2019,2020?

Sincerely Sumit Kumar Ghosh

[Walter Roberson]

http://www.mathworks.co.uk/matlabcentral/answers/43073-a-guide-to-tags

Answer 1

[Greg Heath]

The data set doesn't have the information for predicting anything specific.

Predictions are based on a history of variations of one or more quantities. You only have 1 point for each quantity.

[Sumit Kumar Ghosh]

Sir,

I have made cell matrix for the 10 locations. In the every cell matrix there was 20*1 matrix of parameters. These are input. Then I have made a row of cell matrix where there were assumption of those 20 parameters for 10 locations`s .For every location of this row there were again 20*1 matrix of parameters. This is target. Then I use NARX and specify the input and target. After some training I got output of one row but instead of 10 columns there were 9 columns on that row.

Now my questions are: 1. If the inputs were for 2017, was the output for 2018? 2. I got 9 column for 9 locations but I need for 10 locations. But how? 3. Can I predict for 2020 directly from the input of 2017?