I am having a digitized shoreline of previous years and i want to predict the future shoreline change. please suggest me methods to do so.
You have to change your data as per neural network language. It will analyse the trend of changes for previous years and than it will give future vision based on these experiences.
You can model the shoreline changes based on the currents and tides. If you can do a hindcast of these two variables for the next year, then you can predict how the shoreline is going to change.
There are numerous algorithms:
Consider that, your prediction algorithm is as simple as y=a*x+b which is a linear function.
where:
x is previous values and y will be your prediction value.
You have a and b as unknown parameters. This equation can be solved with at least two values.
If you are required to assess the accuracy, then you need three values.
Follow these steps:
1. Place for example year 2016 values as x and values of year 2017 as y; to solve the function (at least two values).
2. When a and b are calculated ( for example you have y=2x+3.4); by placing previous years values in x, you will get the prediction values for 2017.
Some examples for your prediction model or function will be:
1. k(x,y)=x^t * y (linear kernel)
2.k(x,y)=(x^t * y)^2 (quadratic kernel)
3.k(x,y)=(x^t*y)^d (polynomial kernel)
4. k(x,y)= exp[-omega ||x-y||^2] (radial basis function )
Hopefully, it was clear.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing