Fermé

Developing similar forecasting result of attached paper using LS-SVM code

Developing similar result of attached paper using LS-SVM code

Hi ,

I have some river streamflow data with varied length (River data attached). Some streamflow gauges value may start from 1913 to 2018 and some might start from 1971 to 2018. Total number of streamflow stations are 7.

The data are arranged in monthly format.

At the same time I have some climate index data (total 5 climate index).Climate Data attached. These data are also in monthly format starting from 1900 to 2018.

What I want I want to make monthly forecasting of forecasting as attached paper using LS-SVM model (grid based search) (reference paper attached (figure 4&5) . The LS-SVM draft code is available in following website [login to view URL] They are all in matlab format. The train: test ratio should be 70:30 and 50:50.

Work also include: 1) provide working code using LS-SVM as single file so that it can be run from my laptop Matlab version. I am using Matlab 2019a (academic version) and don’t yet any toolbox setup for LS-SVM.

The expected graphs as below (exemplary graph provided); like figure 4&5 of attached paper):

a) Observed vs simulated against all months one vs one (one streamflow station vs one climate drivers)

:

[login to view URL]

b) Various error calculation as below NSE, Pbias, RE, RMSE to understand skill of forecasted one vs one (one streamflow station vs one climate drivers)

[login to view URL]

c) Box-plot (for observed vs forcasted) against; lines are 25%, 50% and 75%

[login to view URL]

d) Overall skill one vs one (one streamflow station vs one climate drivers). The red line marker is skill for 10%.

[login to view URL]

Please let me know. We can discuss on quotes.

Regards

Compétences : MATLAB

Nº du projet : #24602594