N.K. Zarubina - Lecturer, Southwest State University (Kursk). E-mail: nkzarubina@yandex.ru O.V. Ovchinkin - Ph.D. (Eng.), Southwest State University (Kursk). E-mail: ovchinkin_o_v@mail.ru A.I. Pykhtin - Ph.D. (Eng.), Southwest State University (Kursk). E-mail: sephiroth_kstu@mail.ru

Planning contingent of students is the initial stage of the admission campaign. Formation of a contingent of students in Russia in modern conditions is carried out in accordance with the state of reference for preparation of bachelors, masters, specialists and graduate students, the volumes of which are determined by the Ministry of Education and Science of Russia. In Soviet times, the reception quality was characterized by the successful passing of entrance examinations by the applicant, and the quality of training was determined by the fact of receiving a diploma of higher education. Currently, the reception quality is estimated in most cases such a controversial measure of the effectiveness of how the exam average score of students enrolled in the 1st year. Common tools for assessing the quality of training of graduates are currently not, but one approach is to analyze the percentage of graduate employment and wages. At the same time, universities carry out training undergraduate and graduate students at the expense of physical and (or) legal entities. The number of such students is determined by the universities on their own and, as a rule, fully meet the demand among students. The problem boils down to planning contingent classification of specialties and areas of training with a \"successful\" and \"unsuccessful\" set based on the results obtained during previous campaigns receive. This paper analyzes the data in the South-Western State University receiving campaigns of 2011-2015. Since initially unknown number of clusters and their name, it is necessary to conduct exploratory analysis to identify data structures. To do this, we use the Kohonen neural network designed for problems with uncontrolled training. Please use the clustering results obtained by classical methods set specialties on one indicator - the total number of enrolled students. To calculate the distance between the clusters will use the Ward method. Since the range of values be-tween the results of a set of individual professions can be quite large in order to determine the proximity between objects we will use Manhettenskoe distance. For the treatment of incomplete observations using the average for the replacement string. If all the conditions listed above in the cluster analysis are many areas of training / specialties will be divided into 4 clusters, which can be described as: 1) a very successful set; 2) successfully set; 3) average set; 4) unsuccessful. This 4 cluster planning contingent of students remain on the tree no matter how many years the sample is taken. We compare our results with the results of clustering neural network Kohonen. In this case we see a full match and the structure and content of each cluster. Full match clustering two methods is shown only in the simplest case with an initially predictable result. When clustering over a larger number of different indi-cators the content of the cluster is not the same in most cases - Kohonen neural network is more sensitive to the addition of more performance and clearly demonstrates the changes in the structure of clusters. Thus, it was empirically derived optimal number of clusters using a Kohonen neural network: 4-8. Based on these results it was developed a software product that implements Kohonen neural network for classification of specialties in the planning of the number of students. The user enters the desired number of clusters and data in the specialty obtained in the normal course of the company\'s receptionist. On the basis of the training sample program displays which of the clusters include this profession and what other profession it close in performance. The output of the neural network serves as a recommendation to the decision maker, an ad receiving.

 

1. Pykhtin A.I.,Ovchinkin O.V., Zarubina N.K., Mezenceva A.G.Modifikacii algoritma provedenija konkursa pri prieme na programmy vysshego obrazovanija v sootvetstvii s izmenenijami porjadka priema v vuzy Rossii v 2015 godu // Izvestija JUZGU. Ser. Upravlenie, vychislitelnaja tekhnika, informatika. Medicinskoe priborostroenie. 2015. № 4 (17). S. 45-50.
2. Pykhtin, A. Klevtsova M., Ovchinkin O., Zeveleva I. The concept of innovative system of enrollment in state universities of Russia // Mediterranean Journal of Social Sciences. 2015. T.6. №5. S1. P. 149-153.
3. Zarubina N.K., Mezenceva A.G., Ovchinkin O.V., Pykhtin A.I. Problemy planirovanija kontingenta studentov pri prieme v vuz // ZHurnal nauchnykh i prikladnykh issledovanijj. 2015. № 10. S. 82-84.
4. Pykhtin A.I. Nedostatki ispolzovanija srednego balla EGEH dlja ocenki ehffektivnosti vuza pri prieme na pervyjj kurs obuchenija // Teoreticheskie i prikladnye aspekty sovremennojj nauki. 2015. № 8-3. S. 86-87.
5. Butorin V.M., Anikina E.I., Bochanova N.N., Pavlova E.V. Analiticheskijj obzor informacionno -obrazovatelnykh sistem // Izv. JUZGU. 2012. № 2-3. S. 24-27.
6. Pykhtin A.I.Algoritm prognozirovanija prokhodnogo balla pri prieme v vuz na pervyjj kurs obuchenija // Razvitie nauki i obrazovanija v sovremennom mire: Sb. nauch. tr. po mater. Mezhd. nauch.-prakt. konf. v 6 chastjakh. OOO «AR-Konsalt». 2015. S. 67-68.
7. Zarubina N.K. Ovchinkin O.V., Pykhtin A.I. Ocenka kachestva podgotovki abiturientov pri prieme v vuz po rezultatam EGEH i itogovomu sochineniju // Fundamentalnye issledovanija. 2015. № 8-3. S. 481-485.
8. Zarubina N.K.,Pykhtin A.I., Ovchinkin O.V. Ocenka slozhnosti poluchenija kontrolnykh cifr priema vuzom // Materialy VI Mezhd. molodezh. nauch. konf. «Molodezh i KHKHI vek - 2015». V 4-kh tomakh. Kursk. 2016. S. 48-53.
9. Pykhtin A.I.Emeljanov I.P. Problemy sushhestvujushhikh rejjtingov vysshikh uchebnykh zavedenijj // Materialy Mezhdunar. nauch.-prakt. konf. «Novye informacionnye tekhnologii v obrazovanii». 2011. S. 222-223.
10. Zarubina N.K., Ovchinkin O.V., Pykhtin A.I. Analiz faktorov (kriteriev), vlijajushhikh na pozicii vuza v rejjtinge obrazovatelnykh uchrezhdenijj // Materialy I Vseros. nauch. konf. (s mezhdunar. uchastiem) «Kommunikativnye i obrazovatelnye vozmozhnosti sovremennykh tekhnologijj». Ekaterinburg, 10 marta 2016 g. IOC «Infometod». Ekaterinburg. 2016. S. 27-31.
11. Serebrovskijj V.V.,Starkov E.F., Pavlova E.V., Krasnokutskaja L.N. Razrabotka matematicheskojj modeli peredachi kontenta na osnove sistemnogo analiza postroenija informacionno-obrazovatelnogo multimedijjnogo interaktivnogo prostranstva vuza // Izvestija JUZGU. Ser. Upravlenie, vychislitelnaja tekhnika, informatika. Medicinskoe priborostroenie. 2014. № 2. S. 90-97.