Bryansk, Russian Federation
GRNTI 27.43 Теория вероятностей и математическая статистика
BBK 221 Математика
TBK 6117 Теория вероятностей. Математическая статистика
The work of most information systems involves the processing of data, its accumulation during operation and subsequent analysis. However, the analysis of such a large amount of information by a person is impossible without its preliminary automatic processing. For this purpose, Data Mining is used, which includes descriptive and predictive modeling. The statistical classification is one of the most understandable data analysis technologies for humans and relates to predictive modeling. This task consists in dividing the set of observations into classes based on their formal description. One of the methods for solving the classification problem is logistic regression, while scoring is a common area of application. This article discusses the application of scoring to the problem of assessing the probability of students' expulsion from the University based on data on their attendance and academic performance. The solution of this problem will allow curators of groups, directions and other interested parties to identify the tendency to expulsion in time, identify a risk group among students and take early measures to prevent the event predicted by the built model from becoming a fact. The built scoring model is subject to publication as a web service for further use in the software package for supporting the work of a University teacher. In this case, the model input receives aggregated characteristics obtained from accumulated data on student performance and attendance by the software package, which results in an integrated indicator of the probability of an event, namely, deductions. As a result of building a scoring model, a subsequent assessment of its quality is performed.
Data Mining, statistical classification, scoring, students’ performance and attendance analysis, analytical platform
1. Kalevko V.V., Lagerev D.G., Podvesovskiy A.G. Programmnyy kompleks «Avtomatizirovannoe rabochee mesto prepodavatelya» // Sbornik nauch. trudov II Mezhdunarodnoy nauch. konferencii i XII Mezhdunarodnoy nauch.-prakt. konf. «Sovremennye informacionnye tehnologii i IT-obrazovanie» 24-26 noyabrya 2017 g. M.: Laboratoriya otkrytyh informacionnyh tehnologiy fakul'teta VMK MGU im. M. V. Lomonosova, 2017. S. 197-205. [Elektronnyy resurs]. – Rezhim dostupa: https://www.elibrary.ru/item.asp?id=32661960.
2. Paklin N.B. Optimal'noe kvantovanie dlya povysheniya kachestva binarnyh klassifikatorov // Iskusstvennyy intellekt. – 2013. – V 4. – S. 392-399.
3. Hosmer D. W., Lemeshow S. Applied Logistic Regression (2nd Edition) // Wiley Publishing, Inc., 2000.
4. Kochetkova V.V., Efremova K.D. Obzor metodov kreditnogo skoringa // Juvenis Scientia. – 2017. – № 6. – S.22-25.
5. Analiticheskaya platforma «Loginom» [Elektronnyy resurs]. – Rezhim dostupa: https://loginom.ru/.
