weka Mistakes easily made when using KnowledgeFlow
Introduction
Weka KnowledgeFlow(KF) is a great interface to use. However, Weka manual does not cover every little details of using KF. Here would be a place for collecting those little tricks or details I learnt from those errors I did or will make as time goes. Many thanks to people at Wekalist (especially Mark Hall, Eibe Frank) for building a wonderful learning environment for Weka!
Remarks
TrainingSetMaker and TestSetMaker
- a
ClassAssignermust be linked betweenArffLoaderandTrainingSetMakerorTestSetMaker.
ArffSaver
- In order to save dataset into arff file successfully, it is safer to set
relationNameForFilenameto False inside configuration ofArffSaver.
How to use TimeSeriesForecasting in KnowledgeFlow?
- Open knowledgeFlow, load dataset with ArffLoader
- go to setting, check time series forecasting perspective, right-click ArffLoader to send to all perspective
- go to time series forecasting perspective to set up a model
- run the model and copy the model to clipboard
- ctrl + v, and click to paste model to Data mining process canvas
- save prediction along with original data with ArffSaver
