What is KnowledgeMiner Software?
For more than 15 years KnowledgeMiner Software creates knowledge mining
tools that enable anyone to use data modeling to quickly visualize new
possibilities. It has been doing research and consulting in several fields.
One of these fields it has been active in more intensely is about modeling and
prediction of toxicological and ecotoxicological
hazards and risks of chemical compounds also for regulatory purposes with the goal to
substitute and minimize still widely used animal tests by computer models. Another project
which has started recently and which exclusively and extensively
uses KnowledgeMiner Software tools is focused around climate change related problems,
modeling, and prediction. Sales and demand predictions, macro- and micro-economic
modeling problems, budget and resource planning, energy consumption analysis and
prediction, medical diagnosis of diseases, traffic prediction, wastewater reuse problems
are further examples where KnowledgeMiner tools have been applied to.
Why choose KnowlegeMiner?
The main advantages of the
inductive KnowledgeMiner approach are:
- Only minimal, uncertain a priori information about the
system is required. That means, even if you have no experience
in modeling, data analysis or designing a neural network
you will be able to model, analyze and predict complex relationships
of nearly any kind of system.
- A very fast and effective learning process for a personal
computer. That means you can solve problems on your desktop
in a reasonable time which you may have never thought possible.
- Modeling short and noisy data samples. That means, you
can deal with a problem as is and don't have to construct
artificial conditions for your modeling method to get it
- Output of an optimally complex model. Generally you can
be sure to get a model at the end of the automated modeling
process which can be expected not to be overfitted. Overfitted
models are not able to predict inherent relationships between
- Output of an analytical model as a transparent explanation
component. That means, you can evaluate the analytical model
to explain the obtained results immediately after modeling.
|About KnowledgeMiner® (yX)
Information is not knowledge. - Albert Einstein
What is KnowledgeMiner (yX) for Excel?
KnowledgeMiner (yX) for Excel is a knowledge mining tool that works with data
stored in Microsoft Excel for building predictive and descriptive models
from this data autonomously and easily. It supports all major releases of
Microsoft Excel, 2011, 2008 and 2004. The modeling engine of KnowledgeMiner (yX)
for Excel implements unique modeling technologies which are built on the
principles of self-organization: Learning from noisy data an unknown
relationship between output and input of any given system in an evolutionary
way from a very simple model to an optimal complex one which generalizes well.
KnowledgeMiner (yX) for Excel implements a completely redesigned and
redeveloped modeling engine called (yX). It is based on the modeling
technologies also found in our classic KnowledgeMiner 6.0 software, which
has been successfully used by our customers for more than 10 years.
KnowledgeMiner (yX) in brief
- Unique self-organizing modeling technologies implemented in
the cross-platform (yX) modeling engine (Mac, Windows, Linux) to
make modeling most objective and easy-to-use. More ...
- An analytical model
that describes the data is produced and availabe to the user for
implementing the model in Excel, for example.
Analytical model implemented in a new Excel worksheet by (yX) for Excel.
- High-dimensional modeling which scales to data sets of very
small (< 20) to large number of samples and from small to
large number of potential inputs (<= 50,000). Not any prior
variables selection is necessary! Go self-organize models
from high-dimensional data directly and support our
climate change modeling and prediction project.
- Original approaches to noise immunity and evaluation of models
derived from data to improve reliability of models and to minimize
the risk of getting invalid or chance models from badly sized data sets. More ...
- 64-bit parallel software. Ultra-fast
processing of very computation intensive algorithms by 64-bit,
multi-core/multi-processor, and vector processing support. The
software automatically scales to the number of CPU-cores found
at runtime to take full advantage of current and future multi-core
hardware. The basic rule: Modeling speed grows with every
core and/or processor available at runtime. More ...
- High-end professional modeling software at a very competetive
price-performance ratio without regular update costs. Profit from
future updates free. Prices ...
KnowledgeMiner (yX) for Excel is 64-bit parallel software. It employs a
twofold, simultanious parallelism: Vector processing (single instruction
- multiple data parallelism (SIMD)) and multi-processing (multiple instruction
- multiple data parallelism (MIMD)) to take full advantage of multi-processor and/or multi-core based
Macs. Also, it automatically scales to the number of processing elements found at
runtime (fig. 1).
Fig. 1: KnowledgeMiner (yX) for Excel running at optimal
speed on dual-core MacBooks and iMacs and/or eight-core Mac Pros.
This means, it runs at optimal speed no matter if the machine it is running on is still a single processor
machine or if it is driven by a dual-core or a quad-core CPU or by two
quad-core CPUs, for example (fig. 2). This implies that KnowledgeMiner (yX)
for Excel will also scale to future many-core hardware
not available today. The concluding rule is that (yX) for Excel -
unlike the vast majority of any kind of software on the market today -
runs faster, almost linear, actually, with increasing number of processing
elements and clock speeds (fig. 2).
Fig. 2: KnowledgeMiner (yX) for Excel scales very well to the
number of processor cores actually available at runtime.
The new 64-bit parallel modeling engine results in breathtaking
performance gains. KnowledgeMiner (yX) for Excel
runs more than 600 times faster than the our traditional KnowledgeMiner 6
software as shown in figure 3, according to a recent performance study.
This makes KnowledgeMiner users even more productive
and gets complex modeling and prediction tasks done in almost no time.
Fig. 3: Average speedup of KnowledgeMiner (yX) for Excel on an 8-core
64-bit Intel-based Mac Pro compared to KnowledgeMiner 5.4 running the
same modeling tasks on the same machine.
Starting from version 2.0, KnowledgeMiner (yX) for Excel is high-dimensional
modeling software. It is high-dimensional in both data set dimensions:
It works on data sets with a large number of samples and on data sets with a large
number of potential input variables. Furthermore, the software scales from very small to
large data sets in both dimensions so that the user don't have much to care
about reliability of the resulting self-organized models. Table 1 shows
the ranges which KnowledgeMiner (yX) for Excel is designed
to work in as compared to other common data mining methods.
Table 1: Common modeling technologies and their applicability
to different data set dimensions.
Sample Size n
number of inputs m
(m < 50)
number of inputs m
(49 < m < 500)
number of inputs m
(499 < m < 50K)
(n < 30)
(29 < n < 200)
(yX), [R, NN]
(199 < n < 10K)
(yX), R, NN
(yX), [R, NN]
(10K < n < 1M)
(yX), R, NN
(yX), [R, NN]
NN - Neural Networks and other Machine Learning methods.
R - Statistical regression methods.
(yX) - Self-organizing, high-dimensional modeling in KnowledgeMiner (yX).
[.] - Can be applied under certain conditions only.
A major concern for models built from any data-driven modeling technology
is model reliability, and the risk of obtaining an invalid model grows fast
with increasing number of input variables and with increasing model complexity. Therefore, original research into
noise immunity of high-dimensional state space modeling has been performed
by KnowledgeMiner Software for many years. This research resulted in highly
improved noise immunity algorithms compared to traditional
modeling approaches (see fig. 3). These new algorithms has been implemented
in KnowledgeMiner (yX) for the first time and they additionally allow
at the same time a new model evaluation approach, which helps and supports
the user in a powerful way to assess obtained models. This concept is
described in more detail in the section "Noise Immunity and Descriptive Power".
Based on our independent, cross-platform (yX) core modeling engine,
self-organizing, parallel, high-dimensional modeling with on-the-fly
model evaluation provides a unique and original tool which allows also
non-experts to build reliable predictive analytical models of complex
systems from the desktop with unprecedented power, ease-of-use, and
An example which illustrates the power of KnowledgeMiner (yX) for Excel
is recent studies on modeling and prediction of climate change in
global and regional scales.
One model, for instance, self-organized from 1900 monthly historical air and
sea surface temperature data and over 9300 potential input variables (by using a
time lag of up to 518 months) is composed of rarely 18 relevant inputs, and it took
less than 9 minutes on an 8-core Mac Pro to get this model
(fig. 5)! This
- 36 minutes runtime on a dual-core machine,
- 72 minutes when running serially, all in 64-bit mode,
- about 90 minutes serial runtime in 32-bit space, and
- 3 days and 20 hours if it was running using
our classical KnowledgeMiner 6 software!
- Weeks of work if theory-based and traditional statistical methods are used,
and there are many of such models to develop!
Fig. 5: Global temperature prediction model
self-organized from a high-dimensional data set.
- Mac OS X 10.5 (Leopard) or newer
- Any multi-core Intel-based Mac to run software concurrently (recommended), PPC Macs to run software sequentially, only
- Microsoft Excel 2011, 2008 or 2004