Russian Federation
The article examines the inefficiency of traditional reactive approaches to office cost management, leading to excess inventory or shortage of materials. An algorithm for intelligent consumption forecasting based on machine learning methods is proposed that takes into account a range of internal and external factors: historical data, business unit activity, projects, employee attendance, seasonality, and calendar features. The process of data integration, feature engineering, model selection (gradient boosting, LSTM) and their training is described. The architecture of the algorithm implementation in the existing corporate system in the form of a separate service component with feedback and monitoring mechanisms is considered. The implementation of the proposed approach makes it possible to reduce warehouse surpluses, minimize deficit risks, and increase transparency in budget management.
predictive analytics, consumption forecasting, machine learning, cost management, office expenses, gradient boosting, LSTM, corporate information systems, inventory optimization, intelligent algorithms
1. Kalach, A. V. O vozmozhnostyah primeneniya metoda pochti-periodicheskogo analiza dlya obrabotki izobrazheniy / A. V. Kalach, A. A. Paramonov, S. L. Saharov // Modelirovanie sistem i processov. – 2024. – T. 17, № 3. – S. 44-52. – DOIhttps://doi.org/10.12737/2219-0767-2024-42-50. – EDN HKBSAV.
2. Os'kin, A. F. Primenenie algoritmov intellektual'nogo analiza obrazovatel'nyh dannyh dlya prognozirovaniya akademicheskoy uspevaemosti studentov-istorikov / A. F. Os'kin, D. A. Os'kin // Informacionnyy byulleten' associacii Istoriya i komp'yuter. – 2016. – № 45. – S. 239-240. – EDN ZWOBFH.
3. Aynakulov, Zh. Zh. Intellektual'nyy algoritm ocenki i prognozirovaniya povedeniya slozhnogo ob'ekta / Zh. Zh. Aynakulov, G. E. Kurmankulova // Aktual'nye voprosy sovremennoy nauki : Sbornik statey po materialam XI mezhdunarodnoy nauchno-prakticheskoy konferencii. V 2-h chastyah, Tomsk, 24 aprelya 2018 goda. Tom Chast' 1. – Tomsk: Obschestvo s ogranichennoy otvetstvennost'yu Dendra, 2018. – S. 61-73. – EDN UWTEYJ.
4. Bova, V. V. Prognozirovanie v intellektual'nyh sistemah-assistentah na osnove algoritma poiska kosyakom ryb / V. V. Bova, E. V. Kuliev, S. I. Rodzin // Izvestiya YuFU. Tehnicheskie nauki. – 2019. – № 2(204). – S. 34-47. – DOIhttps://doi.org/10.23683/2311-3103-2019-2-34-48. – EDN MCEWYT.
5. Tehnologiya razrabotki RTL-modeli opisaniya izdeliya pri razrabotke programmno-analiticheskogo kompleksa SAPR / D. V. Shehovcov, A. M. Plotnikov, K. V. Zol'nikov, A. I. Zarevich // Modelirovanie sistem i processov. – 2023. – T. 16, № 3. – S. 79–86. DOI: https://doi.org/10.12737/2219-0767-2023-16-3-79-86
