Options
"performance questions"
Contributor I
Hello. My question involves the speed at which RapidMiner scores(classifies) new examples. Please allow me to provide a bit of context. I have a collection of previously trained and tested classifiers(mostly decision trees and Bayes nets) persisted in a database. As I receive new, unclassified examples, the following process executes.
1) retrieve from the database the appropriate classifier for the incoming example
2) spawn a new command line
3) call RapidMiner from the command line, passing the appropriate process configuration xml, the classifier, and the example
4) collect the results of the classification
On average, the entire process takes approximately 5 seconds to complete for each new incoming example. As I would like to process the incoming examples as quickly as possible, how might I alter the current scheme to achieve faster performance? Thanks for considering my question.
1) retrieve from the database the appropriate classifier for the incoming example
2) spawn a new command line
3) call RapidMiner from the command line, passing the appropriate process configuration xml, the classifier, and the example
4) collect the results of the classification
On average, the entire process takes approximately 5 seconds to complete for each new incoming example. As I would like to process the incoming examples as quickly as possible, how might I alter the current scheme to achieve faster performance? Thanks for considering my question.
Tagged:
0
Answers
I think you might find this link helpful, hope so..
http://rapid-i.com/rapidforum/index.php/topic,331.msg1296.html#msg1296