| Sažetak | U ovom doktorskom radu je prikazan postupak modeliranja informacijskog sustava za potporu odlučivanju u pripremi sportaša. Izrada kvalitetnog programa za pripreme je tek prvi korak u procesu sustavnog praćenja sportaša tijekom trenažnog procesa i u njegovoj ciljanoj izvedbi. Kreiranjem odgovarajućeg podatkovnog modela, te primjenom modernih ICT tehnologija, razvijen je prototip složenog Web sustava, koji na osnovu trenutnog potencijala pojedinog sportaša izračunava njegov maksimalni aerobni kapacitet (VO2max faktor), te kreira odgovarajući program za pripreme. Razvijene su metode i algoritmi za procjenu kvalitete i kvantitete zadanih i ostvarenih treninga. Izračunati trening ekvivalenti su omogućili autonomnu evaluaciju trenažnog procesa od strane razvijenog sustava. Primjenom raspoloživih telematičkih kanala, realiziran je unos povratnih informacija u bazu i skladište podataka ovog sustava u realnom vremenu, čime je omogućena dubinska i OLAP analiza ostvarenih rezultata. Pravovremenim otkrivanjem simptoma pod(pre) treniranosti, dinamički se modificiraju trenažni procesi u realnom vremenu, sa svrhom bolje izvedbe u ciljanom takmičenju, uz smanjenje vjerojatnosti pojave povrede, bolesti ili neadekvatnog rezultata. Razvijene metode i algoritmi su verificirani na odabranom uzorku atletičara dugoprugaša (maratonaca), gdje su pokazali dobar potencijal za širu praktičnu primjenu. |
| Sažetak (engleski) | In times of general and informatical globalization, survival, competitiveness and success on the market directly depends on fast access to proper information. Likewise, the realization of many requirements imposed by modern sport is impossible or incomplete without the use of all available information, knowledge and analyses in real time. In the early beginnings of development of sports IT systems, the importance of storing history data has been recognized, in order to enable the analysis of monitored group of users (athletes) in the past. Since classic transaction systems do not allow such analyzes, it was necessary to develop a data warehouse, where certain methods of data mining and OLAP (On Line Analitical Processing) analysis would significantly expand the possibilities for analysis of their effects, for the purpose of achieving better results in the future bouts. Modern ICT technologies have enabled the realization of ETL (Extracting, Transforming, Loading) processes in real time, thereby allowing the analysis of available information in any instant of the training process of an individual or larger group of users. It is very difficult to develop an information system that would satisfy personal needs of every user while retaining the planned flexibility and modularity. The aim of this paper was to model an information system for support in decision-making and preparations of athletes by using modern information communication technologies with a high level of availability in any moment and in any place. Developed information system had to be able to independently generate a program for preparations of athletes (marathon runners) and to enable them to choose the most appropriate way and moment of communication with the system. At their disposal is developed Web, E-mail or SMS module, by which they can utilize all or some of the available telecommunication channels, according to the situation they are in. Generating the quality program for a specific athletic discipline, or a race, represents only the first step in a systematic monitoring of an athlete in their training process. There should not be the end of interaction between the user and the system. It is necessary to develop methods and algorithms that would, based on calculated potential of every user and their maximal aerobic capacity (VO2max factor), generate program for preparations, evaluate given and done trainings and dynamically evaluate the training process. Recognizing the symptoms of overtraining, undertraining, or inadequate results should lead to correction of certain parts of the generated program, for the purpose of decreasing the possibility of sport injury or illness during the training process, in order to improve their results in a key race. Developed methods and algorithms had to be verified on an appropriate sample of longdistance runners, to test their potentials for a wider use. The introduction shows the idea which was developed by the author in his research. It shows the flow chart of the system`s prototype, which allows the users to generate programs for preparations and the analysis of their results in all micro cycles of the training process. The second chapter describes aims and hypotheses of the research, while the third chapter gives an overview of existing systems and methods. The analysis of existing systems and their disadvantages has inspired and motivated the author to develop the complete system that could independently generate an individualized program for preparations of long-distance runners. The emphasis was on the need to create the system that has the ability of selfcorrection of its parameters and evaluation of the training elements (self-learning), while simultaneously using the available ICT channels, for the purpose of two-way transfer of information in real time. The fourth chapter gives a review of the current usage of ICT in sports. It describes some of the possibilities of implementing algorithms and telematic functions in raising the sport efficiency. The fifth chapter shows the possibilities of using telecommunication resources for the transfer of information in real time. It describes the developed telecommunication subsystem of ORT (On-line Running Trainer) system. The sixth chapter shows the development of the program for preparations of long-distance runners. It explains in detail the algorithms for determining the current potential of long-distance runners, describes the basic principles of trainings and the algorithms for calculating the training coefficients of given and done trainings. Seventh and eighth chapter of this paper show the modeling of the database and the data warehouse, used in the developed system for support in decision-making in the athletic preparations. They also show the periodic ETL process and the solution for logical and physical separation of the transaction database and the data warehouse. The ninth chapter describes used methods of data mining and OLAP analysis. It describes the use of artificial neural networks, for the purpose of recognizing symptoms of overtraining or undertraining. It shows the process of classification and clustering and the possibility of using OLAP analysis to modify the training coefficients and the modification of evaluation of the training elements. The tenth chapter shows all modules of the developed ORT system, with the description of all functions and modes of their usage. The results of the author`s research are shown in the eleventh chapter. The results of all users who participated in the study were compared with their results in the past, without using ORT system. Along statistical data analysis, the results were also verified by using the methods of expert evaluation. The conclusions that were drawn from this multiannual research are shown in the final twelfth chapter. After many years of research, introducing ICT technologies into modern sport training processes has proven to be reasonable. The research led to developing a system which is able to independently generate a program for preparations that is adjusted for every user, a system that enables two-way communication with all users in real time, with the evaluation of given and done trainings in the whole training process. Accumulated knowledge in the database and the data warehouse, with the developed algorithms for the analysis of improvement of every user, will, over time, enable generating even better programs and methods for training evaluation. This is a very complex area that is evolving rapidly and the future will bring the application of many new ideas and methods. The modeled database and the data warehouse are loading in real time by using telematics channels, and the system was dimensioned for a simultaneous interactive use by a large group of users. Such data warehouse and the implemented methods of data mining and OLAP analysis in it, enable the analysis of monitored athletes in the present and in the past, which contributes to the quantity and quality of this expert system for support in decision-making in the athletic preparations. The verification of algorithms and methods was conducted on a limited sample of real athletes (marathon runners). The process of verification realized loading of the database and the data warehouse with quality information, which allowed the modeled system to “learn”, and to permanently evaluate and improve implemented algorithms for generating programs and result evaluation. The results that were achieved by the users show that it is possible to, in such manner, achieve expected (or better) results in a single training process that lasts several months. With the assistance of realized system, it is possible to achieve the results in one season that are better than previous attempts done by the analyzed users. The probability of achieving inadequate results or an occurrence of sport injury is significantly decreased, while the probability of achieving imagined results in the suitable conditions is increased. Modeled system, with appropriate modifications, may be adjusted for some other sport or business area, which opens the possibilities of connection to other similar systems, with the development of new algorithms of wider specter of feasibility. |
