Download
1 / 41
420 likes | 581 Views
The Knowledge Flow Interface. 資管碩一 602630559 段怡安. 12.1 Getting Started 12.2 Knowledge Flow Components 12.3 Configuring and Connecting the Components 12.4 Incremental Learning. Introduction.
E N D
The Knowledge Flow Interface 資管碩一 602630559 段怡安
12.1 Getting Started • 12.2 Knowledge Flow Components • 12.3 Configuring and Connecting the Components • 12.4 Incremental Learning
Introduction • In the knowledge flow users select Weka components from a toolbar, place them on a layout canvas, and connect them into a directed graph that processes and analyzes data. • It helps in visualizing the flow of data
Basic operations • WekaGUI Chooser window • Select the third button i.e. Knowledge Flow Button.
Getting started • Demonstration We will build an ARFF loader that performs a cross validation using J48 • Steps: 1. From the Data Source tab select an ARFF loader and configure it. 2. To specify which attribute is the class we use ClassAssigner object from Evaluation tab.
1. From the Data Source tab select an ARFF loader and configure it.
1. From the Data Source tab select an ARFF loader and configure it.
1. From the Data Source tab select an ARFF loader and configure it.
1. From the Data Source tab select an ARFF loader and configure it.
1. From the Data Source tab select an ARFF loader and configure it.
2. To specify which attribute is the class we use ClassAssigner object from Evaluation tab.
3. Connect the DataSource and ClassAssigner by right clicking on DataSource and select dataset from the menu.
4. Choose the class from ClassAssigner by right clicking on it and selecting the Configure option. 5. To use cross validation, select it from Evaluation tab. 6. Connect the output of ClassAssigner to CrossValidationFoldMaker.
6. Connect the output of ClassAssigner to CrossValidationFoldMaker.
7. Now select a J48 from the Classifiers tab and connect. 8. Connect J48 to CrossValidationFoldMaker first by training set and then by test set.
8. Connect J48 to CrossValidationFoldMakerfirst by training set and then by test set.
9. The next step is to select a ClassifierPerformanceEvaluator from the Evaluation tab and connect J48 to it by selecting the batchClassifier entry from the pop-up menu for J48.
9. select a ClassifierPerformanceEvaluator from the Evaluation tab and connect J48 to it by selecting the batchClassifier
9. select a ClassifierPerformanceEvaluator from the Evaluation tab and connect J48 to it by selecting the batchClassifier
10. Finally, from the Visualization toolbar we place a TextViewer component on the canvas. 11. Add a graph viewer and connect it to J48 ‘s graph output to see a graphical representation of the trees produced for each fold of the cross-validation
10. from the Visualization toolbar we place a TextViewer component on the canvas
