Development of the algorithm of optimal-management statistical model construction

Abstract

Abstract.  Purpose.  Development  of  the  algorithm  defining  an  optimal  system  management  methods,  based  on  a  statistical  model which uses observational data. Methodology. Primarily were implemented: the  preliminary processing  of the initial data  and  testing  its  subordination  to  the  normal  distribution  law.  Based  on  pair  correlation  coefficients  analysis,  the estimation  of  the  relationship-closeness indicators was completed. To form a functioning scheme of studied system were found primary correlation  indexes between resultative index and factor indexes. Determing the regression function of the resultative index to factor indexes, the collinearity of factor indexes checked. As a result, two possible system functioning models were received. On the basis of th at, two  statistical models were received. The regression functions of the resultative index to factor indexes were obtained. Were defined the  internal system relations, which are described by 9 regression functions of factor indexes to factor ones. All calculations a nd research were completed using modern IT-technologies. The regression function existence lets to consider an optimal -management problem as a  nonlinear  programming  problem.  With  the  regression  functions  existence,  it  appears  the  possibility  to  construct  the  optimal management problem for studied system. For this purpose, the Excel program capabilities were used.  Findings.  Both systems were studied; the optimization problem was formulated for each one. A comparative analysis was implemented for the both schemes. T he  obtained statistical model reflects the patterns, which are inherent to studied statistical data, with values  of factor indexes that suit to series  of  inequations  in  model. For  another  selection  model  may  be  slightly  different.  If  it’s  necessary  to  determine  the  pat terns, which  would  characterize  all  industry  of  functioning,  the  selection  size  should  be  increased  and  monitoring  must  be  continued. However,  methods,  used  to  statistical  data  processing,  mathematical  models  constructing  and  making  recommendations  on  the optimal management process as the result of the technique, stay unchanged. Should be noted that statistical analysis, and search fo r the optimal solution is carried out in Excel environment using the add -ins "analysis package" and "Solutions search».  Originality. Using the developed algorithm for a detailed study of the company production capacity. Search for the most optimal ways to affect any  of  the  indicators  of  production  activity.  The  algorithm  can  be  used  for  any  statistical  data.  Practical  value.  The  developed  algorithm is suitable for its further implementation in the management of production of different capacities, the idea and fo undation for software design.