Scribd
Upload
9 views

5 Knime

Uploaded by

tata
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
9 views

5 Knime

Uploaded by

tata
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

Introduction to Knime1

Marta Arias
[email protected]

Dept. LSI, UPC

Fall 2012

1
Thanks to José L. Balcázar for the slides, they are essentially a copy
from a tutorial he gave.
KNIME, I
Concept

KNIME,[naim],
I data mining and data processing visual, modular
environment;
I open source commercial software:
http://www.knime.com/download-desktop;
KNIME, I
Concept

KNIME,[naim],
I data mining and data processing visual, modular
environment;
I open source commercial software:
http://www.knime.com/download-desktop;
I many smoothly-working extensions under KNIME Labs
and under Community Contributions in tech.knime.org,
including
I Text Processing (filters, stemmers. . . ),
I Indexing and Searching (a port of Lucene),
I a port of Christian Borgelt’s data mining software,
I the Pascal-2 STARK-branded Yacaree Associator (initial
version),
I and many many many more.
KNIME, II
Highlights

Programmed as an Eclipse plug-in:


Basically Java, can integrate scripting through the extensions.
I 8 versions:
I Linux/Windows,
I 32/64 bit architectures,
I user/developer versions;
I no sudo/admin/root privileges necessary (does not access
the Windows registry);
I can be fully installed into a USB pendrive - then it can be
used on any machine (of the correct architecture);
I based on tables, (almost) automatic handling of caching
and secondary storage;
I very easy expansion with additional “nodes” (modules).
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
I drag-and-drop into the Workflow Editor;
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
I drag-and-drop into the Workflow Editor;
3. connect compatible node ports by mouse dragging (triangle
ports are tables, always compatible);
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
I drag-and-drop into the Workflow Editor;
3. connect compatible node ports by mouse dragging (triangle
ports are tables, always compatible);
4. configure the nodes (double-click or right mouse button);
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
I drag-and-drop into the Workflow Editor;
3. connect compatible node ports by mouse dragging (triangle
ports are tables, always compatible);
4. configure the nodes (double-click or right mouse button);
5. and. . . run them!
Data Workflows, I
Concept

EASY!
1. Get an empty workflow into the Workflow Editor;
2. add nodes to it:
I find them in the Node Repository, and
I drag-and-drop into the Workflow Editor;
3. connect compatible node ports by mouse dragging (triangle
ports are tables, always compatible);
4. configure the nodes (double-click or right mouse button);
5. and. . . run them!
6. Alternatively, download or import somebody else’s
ready-made example workflows (coming).
Data Workflows, II
Getting Started: Compare 1-NN vs. decision trees on Iris dataset
I Load data into workflow (File Reader node)
I Color-code classes (Color Manager node)
I Split data into training and test sets (Partitioning node)
I Add Decision Tree Learner and Predictor nodes
I Add scorer to get confusion matrix, inspect and highlight
I Add K Nearest Neighbor node with its Scorer to see performance
Data Workflows, III
Get your hands dirty

Elementary tips and tricks


I Have a look at the “Search” capability of the Node
Repository.
I Double-click in the Node Repository makes the node
appear in the Workflow Editor, and connects it to the
currently selected node, if there is one.
I Other run options:
I Single-run button in panel above executes the selected node,
I run-and-show button opens up the result for you to see it,
I multi-run button launches all runnable nodes.
More information available from..

I Book “Beginner’s Luck”


I Knime’s getting-started documentation
I Knime’s short video tutorials

You might also like