I'm using this site http://scikit-learn.org/stable/datasets/ (subtitle 5.5) to create my custom dataset for performing SVM with scikit. Summary of my day: I basically have no idea what I'm doing.
For my thesis I want to predict stock return direction, i.e. the output of SVM should be 1 (UP) or -1 (DOWN). At the moment I'm trying to figure out SVM with a random sample (because I do get how the tutorials work).
As on the mentioned website it says that each line takes the form <label> <feature-id>:<feature-value> <feature-id>:<feature-value>, I thought that the training set I provide should take the same formatting. Hence I created following training sample in Notepad++:
<1> <1>:<0>, <1>:<19260800>, <1>:<77.83>
<1> <2>:<-1>, <2>:<20110000>, <2>:<75.78>
<-1> <3>:<1>, <3>:<53306400>, <3>:<76.24>
<1> <4>:<0>, <4>:<61293500>, <4>:<78.00>
<-1> <5>:<-1>, <5>:<42649500>, <5>:<75.91>
For example, the second line:
<1> means that stock went up since the day before, <2> is the of the data of the second line, <-1> is a negative Twitter sentiment for that day for the specific firm, <20110000> is the stock volume for that day, <75.78> is the adjusted closing price of that day.
I hope you understand what I'm trying to say. And I hope even more somebody can help me out.
Thanks in advance!
