I am not sure if others have seen this tool, but this is a great free graphically tool for analyzing data-sets. It can handle static data, classification, and time-series data. This is a great tool for Data Mining.
Please register on futures.io to view futures trading content such as post attachment(s), image(s), and screenshot(s).
On the Diagonal above, you can see the classes predicted vs the true result. These are the correct predictions.
Class 2 (chop): 439 Correct
Class 1 (short): 8 Correct
Class 3 (long): 2 Correct
Ideally, I am looking for a automated system where all examples show on the diagonal (every trade called correctly), but I could live with some Class 1 & 3 being called chop (missed opportunity). What would hurt is entering a trade, a 2 being called a 1 or 3 (entering randomly on chop) or a 1 or 3 called a 3 or 1 (long vs short) error. This is what I am out to minimize in my SVM thread.