Glad to help Ron. I'd love to know more about how you achieve your results as well.
When I was getting my degree in AI, Expert Systems were all the rage. Most of us thought Bayesian systems were better than black box learning because these systems could produce explanations of their "thought", which made it easier for experts to validate them, for users to trust them and to know when they might be getting off track. That doesn't matter so much in perception-like tasks, like a lot of black box learning is used for, but for many applications it matters.
When I started thinking about applying what I knew to markets, I decided against NN, random forests, et al. partly for this reason. The other issue I worry about is that such systems inevitably need retraining. The fundamental problem isn't just detecting this need, but also, "What data should be used to retrain?" In developing the initial system you need training data, validation data and test data. Typically, the test data is going to take you right up to the time you put the system in production.
What happens when the system needs retraining? You only have the data from the time it was in production to use for the retraining (or you need to start system development from scratch). How often is that going to be enough data for all of training, validation and test? I don't know but I wasn't eager to find out.
With Bayesian, I can try to express very long lasting -- perhaps fundamental -- market concepts in the Bayesian formulas, and the statistics of how well those concepts relate to specific decisions self-update over time. I then can do walk-forward tests over 10 to 20 years right up to the present to validate the concepts. For this kind of application, I think it is a safer way to work.
__._,_.___
Posted by: rosenberggregg@yahoo.com
| Reply via web post | • | Reply to sender | • | Reply to group | • | Start a New Topic | • | Messages in this topic (29) |
**** IMPORTANT PLEASE READ ****
This group is for the discussion between users only.
This is *NOT* technical support channel.
TO GET TECHNICAL SUPPORT send an e-mail directly to
SUPPORT {at} amibroker.com
TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
http://www.amibroker.com/feedback/
(submissions sent via other channels won't be considered)
For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/
This group is for the discussion between users only.
This is *NOT* technical support channel.
TO GET TECHNICAL SUPPORT send an e-mail directly to
SUPPORT {at} amibroker.com
TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
http://www.amibroker.com/feedback/
(submissions sent via other channels won't be considered)
For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/
.
__,_._,___
EmoticonEmoticon