12/13/2011
Summary
RAPIDMINER
FREE SOFTWARE FOR DATA
MINING, ANALYTICS AND
BUSINESS INTELLIGENCE
RapidMiner project
Strengths
How to use RapidMiner
Operator highlights
RapidMiner GUI
References
DataBase And Data Mining Research Group (DBDMG)
RapidMiner Project
A fully integrated environment for machine learning,
data mining, text mining, predictive analytics and
business intelligence
It is distributed under the AGPL open source license
and has been hosted by SourceForge since 2004
It can be used as a stand-alone application for
data analysis or as a data mining engine for the
integration into own code
Operator highlights (1)
Data mining modeling:
Support Vector Machines (SVM),
Rule learners
Decision trees
Bayes
Gaussian Processes
Neural Networks
Evolutionary optimization
Boosting
Apriori
FPGrowth
Clustering
and many others
How to use RapidMiner
RapidMiner can be used in several ways:
As a standalone tool by means of the simple GUI,
connecting the requested operators to build your process,
executing it and getting its result directly in the RapidMiner
environment
As a batch process one can build the workflow by means of
the GUI and then execute it running the RapidMiner script
with the XML process as input
As a Java API one can integrate the RapidMiner facilities in
your own data mining or business intelligence code building
the requested process directly inside the java code
As an hybrid solution one can build the process with the GUI
to executing and to managing it inside a Java code
Operator highlights (2)
Data Transformations:
Aggregation
Discretization
Normalization
Filter
Sampling
PCA
Missing
Lot
value replenishment
more
12/13/2011
Operator highlights (3)
Evaluation:
Cross-validation
Leave-one-out
Sliding
time windows
testing
Significance tests
ROC
Etc.
Back
Welcome Perspective
Download and launch RapidMiner
Download:
http://sourceforge.net/projects/rapidminer/files/1.%2
0RapidMiner/5.1/rapidminer-5.1.014.zip/download
Launch:
Double click:
rapidminer-5.1.014\rapidminer\lib\rapidminer.jar
Or from command prompt, in the RapidMiner root
directory:
java jar ./lib/rapidminer.jar
Repository (1)
When you launch RapidMiner for the first time, it asks you to create a new
Repository to store/load processes and data.
Repository (2)
You have to set the name of the new Repository and the path of its physical
location on the disk.
RapidMiner GUI
Creating a new process, the GUI generates an XML file that defines the analytical
processes the user wishes to apply to the data. This file is then read by the
RapidMiner engine to run the analyses automatically. While these are running, the
GUI can also be used to interactively control and inspect running processes.
12/13/2011
RapidMiner GUI Perspectives
Design Perspective: is the central RapidMiner
perspective where all analysis processes are
created and managed
Result Perspective: If a process supplies results then
RapidMiner takes you to this Result Perspective
Welcome Perspective: first perspective when
RapidMiner is lunched, where you can see the last
executed processes and some logs.
Design Perspective
In this view all the work steps (called operators) available in RapidMiner are
presents and they are used as building block for every process. The repository
section serves for the management and structuring of your analysis processes
into projects and at the same time as both a source of data as well as of the
associated metadata.
Operators (1)
Working with RapidMiner fundamentally consists in defining analysis process by
indicating a succession of operators.
Expert mode
To set your custom parameters for the models you need to enable the expert
mode.
Design Perspective
The process view shows the individual steps within the analysis process as well as
their connections. New steps can be added to the current process. Connections
between them can de defined and detached.
Operators (2)
The inputs and outputs of operators are generated and consumed by ports. Every
operator is defined by its inputs, outputs, action performed and parameters.
12/13/2011
Operators Load data
Store data into Repository (1)
To store a dataset into
Repository you can use the
Data import Wizard.
You can load data in different ways: from repository, csv, excel, etc..
Store data into Repository (2)
Process Discretization
Process Classification
Process Validation (1)
12/13/2011
Process Validation (2)
Process Association rule extraction
Remember: transactional data in RapidMiner are treated as binominal values
(Item presents: true/false). You can use a pre-processing operator to do this
conversion.
4.2. Displaying Results
4.2 Displaying Results
Process - Clustering
We have already seen t hat object s which are placed at t he result port s at t he
right -hand side of a process are aut omat ically displayed in t he Result Perspect ive
aft er t he process is complet e. T he large area on t he t op left -hand side is used
here, where t he Result Overview is also already displayed, which we will discuss
at t he end of t his chapt er.
Result Perspective
Each current ly opened and indicat ed result is displayed as an addit ional t ab in
t his area:
Figure 4.2: Each open result is displayed as an addit ional t ab in t he large area
Objects which are placed
result
on t he at
t op the
left -hand
side.ports at the right-hand side of a process
are automatically displayed in the Result Perspective after the process is
St rict ly speaking, each result is also a view, which you can move t o anywhere
completed. Each
currently
opened
and
indicated result is displayed as an
you wish as usual. In t his way it is possible t o look at several result s at t he same
You area.
can of course also close individual views, i.e. t abs, by clicking once
additional tabt ime.
in this
on t he cross on t he t ab. T he ot her funct ionalit ies of views, such as maximisat ion
77
Plot View Confusion Matrix
One of the strongest features of RapidMiner are the numerous visualisation
methods for data, other tables, modells and results offered in the Plot View.
Plot View Clustering result
One of the strongest features of RapidMiner are the numerous visualisation
methods for data, other tables, modells and results offered in the Plot View.
12/13/2011
Plot View Complex plot
References
Mierswa, I., Wurst, M., Klinkenberg, R., Scholz, M.
and Euler, T., Yale (now: RapidMiner): Rapid
Prototyping for Complex Data Mining Tasks. In:
Proceedings of the ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining
(KDD 2006)
http://rapid-i.com
One of the strongest features of RapidMiner are the numerous visualisation
methods for data, other tables, modells and results offered in the Plot View.
