01-11-2017 06:56 AM
I have build an algorithm which i need to validate its performance against current modelling tools in rapid miner but am struggling to get this working.
Basically i have a written a script in C# pulling data from a SQL Server. I intend to measure the results of this script against the result of existing data modelling techniques within rapidminer when applied on the same dataset.
Any suggestions please.
01-11-2017 11:03 AM
Does your C# script do the modeling as well? If it only pulls in data from SQL server, you can do all the modeling in RapidMiner by using a Read Database operator to pull in the data.
If your C# script does some analysis, you'll need some sort of performance measures to compare them too in RapidMiner. Are you doing classifcation? Maybe Precision/Recall for the classes, kappa, etc.
01-11-2017 02:58 PM
You could do a Cross Validation in RapidMiner and export the actual training and test sets. Then apply your algorithm on the training set, apply the resulting model on the test set, and compare with the performance of RapidMiner's learners.