Computational Statistics

Computation statistics based software is currently widely used for general and special purpose applications, especially when it is necessary to process massive data flows.

We apply graphic procedures, various types of simulating, resampling, permutations, bootstrapping, Gibbs sampling, all contributing to optimizing conventional Bayesian algorithms and handling huge volumes of data. Such an approach allows for efficiently predicting and result interpreting for solving complicated problems in business, science and technology.


We perform in the following areas:

  • 01Sophisticated statistical models
    We successfully solve statistical model behavior problems by linking simulation algorithms to Monte Carlo algorithms used for computational mathematics problems. That allows constructing convenient computational models and obtaining necessary object parameters.
    To solve a problem, we first develop random number generation programs with the help of appropriate methods and means, then we use these numbers to get random values or random processes with more sophisticated distribution laws, and these enable calculating the model's characteristic values and processing outcomes of the experiment.
    Computing capacity growth and development of corresponding numerical algorithms brought up the keen interest in nonlinear statistical models, namely artificial neural networks. These researches made the foundation for development of sophisticated statistical models like generalized linear model or hierarchical model.

    Applications include researches with an immense number of variables but relatively small number of cases. The analysis of such data reaches the limit of many statistical methods, requiring more sophisticated models. The analysis of such data requires statistical methods with sophisticated statistical models. Personalized medicine applications may serve as an example.
  • 02Trends estimation
    We solve data interpretation problems based on trends statistically discovered in the data. In addition to searching, we use the discovered trends to describe the obtained data.
    The trends may be described by various equations — linear, logarithmic, polynomial, etc. We determine the actual type of the trend by matching its functional model using statistical methods or smoothing the initial time series.
    Trend lines are widely used in the technical analysis. Currently, there is a great number of methods to construct and interpret them. For example, in industrial automation applications (i.e. in a SCADA system) trend may be represented by an event log showing status of each input and output of a PLC with appropriate time stamps. A chart plotted based on such an event log may vividly demonstrate production process dynamics.

    The trends estimation has become commonly used in economics: trends estimation in the credit market, interest rate dynamic estimation, etc. The trends estimation is used in global weather forecasting by approximating long-term observations and drawing conclusions on what future weather might be. The trends estimation – analysis of search requests on specific topics – may be carried out in the Internet.
  • 03Predictive analytics
    Our aim is to get outside the limits of knowing what has already happened and to provide the best estimation of what may happen in future. Our foundation is statistical analysis of acquired data and predicting further behavior of objects and events in order to make an optimal decision. To achieve this goal, we use statistical, data mining and game theory methods.

    Applications may include actuarial computations, financial services, insurance, telecommunications , retail, tourism, health service, pharmaceutics and others. One of widely known applications is loan scoring.

Contact us

If you want more information fill in this form. You will be contacted as soon as possible. Please fill in all required fields.

Get In Touch

  • Bolshoy prospekt V. O. 80R
    194044, St. Petersburg, Russia
  • +78126709095
  • This email address is being protected from spambots. You need JavaScript enabled to view it.