One of KnowledgeMiner's major strengths is that it easily
allows prediction of dynamic systems using both regression models and Similar Patterns technology.
Another advantage of KnowlegeMiner is its knowledge extraction
capabilities. It self-selects the relevant inputs, generates
an optimal composition of them (model), and provides an explanation
component (equation, or patterns) on the fly.
Selected Publications
An open collection of publications related to application of self-organizing data mining technologies in various domains.
Climate
change modeling and prediction project
High-dimensional modeling and prediction of air and sea surface temerature anomalies data till 2020.
SIAM SDM'11 Data Mining Challenge: Prediction of Biological
Properties of Molecules from Chemical Structure (1m - PDF)
The problem was to classify chemical molecules into substrates
and non-substrates of the CYP 2C19 isoform of the cytochrome
P450 enzyme in human. Predicting whether or not a particular chemical will
be metabolized by 2C19, as well as other major isoforms of CYP P450, is of
primary importance to the pharmaceutical industry. This problem description, however, was not
disclosed before the end of the challenge, which was entirely based on numerical data, only.
KnowledgeMiner submitted a top three solution using KnowledgeMiner® (yX)
for Excel right out of the box. May 2011.
Knowledge mining in Life Sciences
About the objective of building mathematical models for toxicity prediction. Article in the
journal Chemistry & Industry, August 2008
Examples that come with KnowledgeMiner
An overview of the examples that come with the KnowledgeMiner
package. Each example file contains ready-to-use models
for quick evaluation and testing. Download
our free trial demo to test out these examples.
Getting Started with KnowledgeMiner
Please check out our beginner's tutorial to get a quick
leg up in using KnowledgeMiner. This tutorial shows how
KnowledgeMiner was used to predict the boiling point of
hydrocarbons.
Teaching KnowledgeMiner to Add
A simple example that shows how KnowledgeMiner can quickly
uncover relationships in three columns of data using the
Input-Output model method.
Modelling and Prediction of Toxicity of Environmental Pollutants (284k - PDF ©Springer-Verlag)
Introduction, problem description, and concept outline for building Quantitative Structure-Activity Relationship (QSAR) models of ecotoxicological compounds like pesticides to reduce animal tests for dossier preparation for regulatory purposes.
First results from a European research project, DEMETRA, presented at KELSI 2004 conference, Milan, Italy.
Contributed by Frank Lemke, 11/29/04
Does my model really reflect a causal relationship? (HTML or PDF - 240k)
The noise filtering power of GMDH, do not believe in a model's closeness-of-fit, only, and introduction of a new model quality measure: Descriptive Power.
Contributed by Frank Lemke, 10/17/02
Knowledge Extraction in Proteomics (40k - PDF)
Over the last several years there has been a tremendous interest in proteomics with significant changes in technology. As the field continues to expand there has become an increasing need to mine large amounts of data... (more)
Contributed by Frank Lemke, 08/25/02
Carcinogenicity
Prediction (Part 2) (36k - PDF)
Carcinogenicity prediction of aromatic compounds from a
set of molecular descriptors using a self-organising data mining approach.
Contributed by Frank Lemke, 08/25/02
Draft for a predictive controlled
financial trading (12k - PDF)
New approach to minimizing risk of financial trading by
applying self-organizing data mining at different levels
of knowledge extraction.
Contributed by Frank Lemke, 07/17/01
Carcinogenicity
Prediction (220k - PDF)
Carcinogenicity prediction of aromatic compounds from a
set of molecular descriptors. (Data set from the COMET project,
ENV4-CT97-0508, funded by the European Commission).
Contributed by Frank Lemke, 06/09/00
Medical Diagnosis
(52k - PDF)
Diagnosis of heart disease using GMDH, Fuzzy Rule Induction
and Analog Complexing on the Long Beach data set (UCI ML
Repository). Also some cost reduction aspects from using
the generated models.
Contributed by Frank Lemke, 06/09/00
Financial Trading System
Performance results of 3 different stocks using a five-day
predicted trading indicator. Trading signals generation;
Tested Period: 07/01/96 to 04/17/98 (454 days).
Contributed by Frank Lemke, 03/22/99
Applications of Adaptive Learning
Networks in the USA prior to 1984
From Farlow's book: Self-Organizing
methods in Modeling. GMDH Type Algorithm.
