数据挖掘（Data Mining）——Pentaho Weka

 关于 Pentaho Weka Experimenter 功能的使用

1. Automate the process of determining the best method to use
2. Is an interactive process in the Explorer or Knowledge Flow
3.  Run classification and regression algorithms on a corpus of data sets
4.  Try different parameter settings
5. Collect performance statistics
6. Perform significance tests on the results
7.  Raw output saved to files or databases
8.   Analysis results can be export as text, CSV, Gnuplot, LaTeX or HTML
9. Advanced users can distribute the processing load across multiple machines

关于 Pentaho Weka Knowledge Flow 功能的使用

1. Define a data mining “process”
2. Like the Explorer, all of WEKA's algorithms are available
3.  Data flows through the process from node to node
4. Accommodates both batch-based processing and data streams
5. Command line interface to WEKA can also train incremental classifiers on data streams
6.   Fully multi-threaded
7. Accommodates multiple independent “flows” on the same layout
8. Knowledge Flow’s Classifier step is multi-threaded
9. Build models for more than one cross-validation fold in parallel
10. Binary and XML-based persistence of flow layouts

关于 Pentaho Weka Explorer 功能的使用

·“Preprocess” panel

·Load data from various sources (file, SQL database, URL etc.)

·Apply pre-processing “filters” to the data

·Summary statistics & histograms

 

·“Classify” panel

·Apply classification and regression algorithms

·Evaluate resulting models

·Numerically via statistical estimation

·Graphically through visualization (data and model)

 

·“Cluster” panel

·Apply clustering algorithms to the data

·Visualize the outcome

·Clusters that represent density estimates can be evaluated based on the statistical likelihood of the data

 

·“Associate” panel

·Learn association rules for market-basket type analysis

 

·“Select attributes” panel

·Mix and match algorithms for evaluating the utility of attributes and sets of attributes with different search methods

 

·“Visualize” panel

·Color-coded scatter plot matrix of the data

·Select, enlarge, zoom in etc.