Deductor

Deductor is the analytical platform, i.e. a basis for creation of the finished application solutions. The technologies implemented in Deductor allow to pass all stages of creation of an analytical system based on uniform architecture: from creation of the data warehouse before automatic selection of models and visualization of the received results.

Deductor provides to analysts the work benches necessary for the solution of the most various analytical tasks: the corporate reporting, forecasting, segmentation, search of patterns – these and other tasks where such analysis techniques as OLAP, Knowledge Discovery in Databases and Data Mining are applied. Deductor is the ideal platform for creation of decision making support systems.

Tasks

The technologies implemented in Deductor can be used as in a complex, and separately for the solution of a broad spectrum of business problems:

• Systems of the corporate reporting. The ready data warehouse and flexible mechanisms of preprocessing, cleaning, loading, visualization allow to create quickly finished reporting systems in a short time.
• Processing of independent requests. The end user can receive with ease the answer to questions like "How many was sales of goods on groups to the Moscow region for last year with breakdown on months?" and to browse results the most convenient for it by method.
• Analysis of trends and patterns, planning, ranging. Ease of use and an intuitive data model allows you to carry out the analysis by the principle "that-if", to correlate your hypotheses to the data which are stored in the database, to find abnormal values, to assess the consequences of acceptance of business solutions.
• Forecasting. Having constructed model on historical examples, you can use it for forecasting of a situation in the future. In process of change of a situation there is no need to rebuild everything, it is necessary to doobuchit model only.
• Risk management. The algorithms implemented in a system give the chance to decide rather precisely on what characteristics of objects and as affect risks thanks to what it is possible to predict approach of a risk event and beforehand to take necessary measures to decrease in the size of possible adverse effects.
• Data analysis of marketing and sociological researches. Analyzing the information about consumers, it is possible to define who is your client and why. As their addictions depending on age, educations, a social status, a material status and a set of other indicators change. Understanding of it will promote the correct positioning of your products and sales promotion.
• Diagnostics. The analysis mechanisms which are available in the Deductor system with success are used in medical diagnostics and diagnostics of the difficult equipment. For example, it is possible to construct model on the basis of data on failures. With its help quickly to contain problems and to find the reasons of failures.
• Detection of objects on the basis of indistinct criteria. Often the situation when it is necessary to detect an object meets, being based not on such accurate criteria as cost, technical characteristics of a product, and on vague formulations, for example, to find the products similar to yours from the point of view of the consumer.

It is only the small list of solvable tasks. Actually it is about any tasks where it is required to consolidate data, to display them by different methods, to construct models and to apply the received models to new data.

Composition

Deductor consists of five parts:

• Studio is the program implementing functions of import, processing, visualization and export of data. Deductor Studio can function also without data warehouse, obtaining information from any other sources, but the most optimal is their sharing. Deductor Studio includes the full range of mechanisms allowing to obtain information from any data source, to carry out all operation cycle (cleaning, transformation of data, creation of models), to display the received results in the most convenient way (OLAP, charts, trees …) and to export results on the party.
• Viewer is a workplace of the end user. Allows to separate process of creation of scenarios from use of already ready models. All difficult operations on preparation of scenarios of processing are executed by analysts experts through Deductor Studio, and Deductor Viewer provides to users an easy way of work with ready results, hides from them all difficulties of creation of models and does not impose high requirements to qualification of employees.
• Warehouse is the multidimensional data warehouse accumulating all information, necessary for domain analysis. Use of uniform storage allows to provide data consistency, their centralized storage and automatically provides all necessary support of process of data analysis.
• Server is the service providing remote analytical data processing. Allows to process automatically data and to retrain models on the server, optimizes accomplishment of scenarios due to caching of projects and use of multithreaded processing.
• Client – the client of access to Deductor Server. Provides access to the server from third-party applications and management of its work.

The architecture implemented in Deductor allows to achieve the maximum flexibility during creation of the finished solution. Thanks to this architecture it is possible to assemble all necessary tools of the analysis in one analytical application and to implement automatic execution of the prepared scenario.

The technology platform includes the means allowing to reduce as much as possible development time to quickly create and bring new application solutions to the market and in short terms to adapt them according to the changing requirements of the enterprises. Possibilities of the platform provide not only the high speed of initial product development, but also its fast adaptation further.

Creation of the finished solution takes not enough time. It is enough to obtain data, to define the scenario of processing and to set the place for export of the received results. Existence of powerful set of mechanisms of processing and visualization allows to move on steps, from the easiest ways of the analysis to more powerful. Thus, the user receives the first results practically at once, but at the same time it is possible to increase solution power easily.

New view on data

Deductor allows to look absolutely in a new way at data, squeezing out of them a maximum of valuable information. It integrates all tools, necessary for the analysis, and gives to users huge opportunities:

• Powerful analytical platform;
• Modern self-training mechanisms of the analysis;
• Uniform data warehouse;
• The single user interface for any mechanisms of the analysis;
• Package accomplishment of scenarios of processing;
• Remote analytical processing;
• Separation of work of the analyst from the end user;

Decision support became an integral part of any modern information system, but a large number of legacy programs and high cost of analytical tools constrains distribution of the systems of this class. Using Deductor this problem is solved. Now thanks to the affordable price of this product the powerful analytical systems are available to most the companies.

2012: End of lifecycle of Deductor

By 2012 it becomes obvious that Deductor ceases to meet requirements of quickly changing data analysis market: explosive growth of number of data, new analysis algorithms, promoting of web services, distribution of smartphones and tablets.

For ensuring revolutionary break the new platform with the web interface, essentially new opportunities of design and reuse of analytical models, the convenient publication of web services was required. All this could not be implemented within the existing architecture. It is decided to develop from scratch the platform of future generation — Deductor 6.

2008: Ready-made solutions of Deductor

In 2008 ready-made solutions appeared:

• Deductor Credit Scorecard Modeler is process automation of creation of scoring cards.
• Deductor Data Quality is ensuring quality of client data.
• Deductor Credit Pipeline is the credit pipeline for operational service of a big flow of clients without quality degradation, with observance of all rules and risk management policies of bank.
• Deductor Demand Planning is purchase planning of goods for the purpose of minimization of deficit and increase in turnover.

2007: Development of Deductor and connection of universities

Release of Deductor 5: significant expansion of opportunities of the previous version regarding increase in quantity of algorithms of processing, visualizers, supports of data warehouses. The structure of the platform includes a server component — Deductor Server.

The affiliate program is developed for higher educational institutions. Universities began to be connected to an educational initiative of BaseGroup Labs. Over time more than hundred universities include Deductor in the official training program.

There is an educational portal. Training is conducted on the basis of the free version of the analytical Deductor Academic platform. On the website articles with the description of methodology of data analysis, practical application of analytical algorithms, a mathematical apparatus are published.

Same year the affiliate program earned. BaseGroup Labs gives all support to the partners: training, certification, consultations, attraction to project implementation, joint marketing, accounting of wishes of partners in development process of new versions of Deductor.

2004: New architecture of Deductor

To release of Deductor 4 there was a change of architecture, as a result, radical changes in a system. Deductor is transformed from the software package to the analytical platform.

There is an opportunity to build scenarios, combining any methods of data processing for obtaining the finished application solution of a business challenge. Deductor 4 allowed to pass all stages of creation of an analytical system based on uniform architecture: from creation of the data warehouse before automatic selection of models and visualization of the received results.

2001: Deductor release

The company made strategic decision — to depart from the ordered development and to concentrate efforts on creation of mass products and solutions in data slicing pane.

Results of developments of previous years were collected in the first release of an application program package of Deductor. Over time applications were a part of a packet:

• Cube Analyzer is data analysis on the basis of OLAP technology;
• Tree Analyzer is data analysis on the basis of decision trees;
• SOMap Analyzer is information analysis on the basis of self-organizing cards of Kokhonen;
• Neural Analyzer is data analysis on the basis of multilayer neural and RBF networks;
• RawData Analyzer is cleaning and predata processing.
• Applications of a packet are integrated among themselves that allows to transmit results of processing from one application to another. Support of the multilingual interface is added.

The first projects in the field of advanced analytics both with Russian, and with the foreign companies are executed.