The goal that inspires our team is to introduce data mining to mass users and provide the most simple, clear, and efficient way to this powerful and intelligent tool. Modern achievements in data visualization and cognitive data processing approach allows almost untrained people to use data mining as regular as calculator, text editor, or web browser.
The successful application of existing techniques of deep analytics requires:
- Understanding the nature of the data and having knowledge on the area of expertise.
- Developed intuition of algorithms application and results interpretation.
The first requirement refers to the domain expert. For the second requirement often professional mathematician is responsible, who is - an expert in the field of machine learning with experience in the specific analytical tools.
We believe that now we have all preconditions in order to shift the focus in this pair to a domain expert and then completely exclude professional mathematicians from the process of analysis and decision making.
The basis of our confidence
First of all it is a new paradigm of data mining, developed and improved by the scientific school of Prof. N.G. Zagoruiko. The approach proposed by Prof. Zagoruiko is based on the assumption that the person solving the problem of finding empirical laws, applies to the universal psycho-physiological mechanism of cognition, evaluating the measure of similarity between objects and looking for maximum of compactness and simplicity of the model description in terms of this similarity measure.
Today's volume of data submitted for analysis eliminates the requirement of using sophisticated algorithms. These algorithms are overloaded with explicit and implicit assumptions about the nature of the hidden patterns in the data that is caused by small amounts of training samples.
In our work, we rely on that it is much more efficient to spend additional unit of time to process a greater volume of information by simpler algorithms than spend it on building a complex model for a small amount of data. This approach has two major advantages:
- "Simple" algorithm is more understandable to the user; hence, the analysis becomes more manageable.
- The result is more reliable as more data is processed.
Everybody solves data mining problems as a regular practice. The generalization of facts, statistical models building, prediction of the environment behavior - all these are the natural process of thinking. That is why we believe in deliverable of our goal to provide all the power of data mining to everyone in a simplest way.
The human should be equipped with the right tool to remove restrictions on the amount of data allowed for direct analysis. Similarly, as the bicycle allows him to drive faster and further due to the use of natural muscle strength