My first post concerns an exiting project I am working with, relating to implementing the SVM on time series data using LIBSVM (code available via LIBSVM -- A Library for Support Vector Machines) in MATLAB.
Though I understand a bit of how SVM works mathematically (I have only taken one course in applied linear optimization, so my competence in this field is limited), I have no idea how to implement it. I have read all the popular tutorials on the subject, alas, they assume you know the basics. The basics include:
1. Choosing features
2. choosing classes
using MATLAB code.
My main question relates to what do to with my data (ex. how to group it into features in order to use the cross-validation function, how to adapt the data to use the svm-train function etc).
Has someone here implmented SVM using MATLAB, and has the time and kindness to help me out?