Competence

Our team of experts is ready to help solve your knowledge mining problems in these primary application fields:

• Sales and demand prediction;
Sample graph >>
• State-of-the-art QSAR modeling for industry, consumer safety, and regulatory bodies;
Sample Excel file: Model, data, and predictions on developmental toxicity of selected chemical compounds. (582 KB) >>
• Development of predictive models for geophysical problems,
Example solutions >>

Knowledge Mining Services
If you have a one-time modeling, classification or prediction problem or if you just want to learn how to get the most out of your data, our knowledge mining service may be what you are looking for. Based on a non-disclosure agreement, if desired, you send us your data and you will get back a solution that:

1. Is composed of relevant inputs. Our algorithms automatically select the important inputs from the unimportant ones during self-organization of models.
2. Describes your data appropriated. Based on the data's noise level, we use the algorithms which are most suited to model the data. Also, a corresponding explanation of the data will be provided in forms of:
   • A set of linear or nonlinear regression models, or
   • Fuzzy/logic IF-THEN rules, or
   • Similar patterns/cases, or
   • Cluster of similar data samples or variables.
3. Generalizes well on new data. Our self-organizing modeling technologies have a built-in mechanism that generate optimal complex models with respect to the data's noise level to avoid overfitting the data. On a most possible likelihood this ensures that the generated models also predict well on new data.
4. Provides prediction uncertainty. Real-world data sets contain noise. Noise means uncertain results. The set of predictions you obtain reflect this uncertainty, individually, in an intuitive way.
5. Is availabe in a competetive short period in time. This lets you quickly start using provided models saving your time and money.

Results and models will be summarized in a report and will become your property. Interpretations, suggestions, and conclusions of the results are not included in the data mining service. Check out full-service consulting for this.

High-dimensional Modeling
Today, lots of data are observed and recorded every day. Increasingly, predictive modeling also means dealing with many potential input variables and with variables selection in an objective way. Generating predictions from mass spectrometric data, molecular properties for QSAR modeling, or from climate data, for example, means finding relevant variables from the amount of recorded ones. If you have such a high-dimensional modeling request of thousands or tenthousands of input variables our unique high-performance parallel 64-bit modeling tools can help you out here. Many smaller looking modeling problems can be high-dimensional, too, if you have more variables than samples in the data, for instance, or if the variables-to-samples ratio is high. Make sure to consider this for getting valid models.

Full-Service Consulting
In addition to our knowledge mining services, you are interested in a longer running project assistance, including data preprocessing, feature extraction, modeling, interpretation of results, implementation of generated models in MS Excel or in C code, or programming requests, then email us for our consulting service.