Napovedovanje denarnih prilivov z algoritmom XGBoost : magistrsko delo

Ognjanović, Špela (author), Grošelj, Jan (mentor), Dular, Tomaž (comentor)

Uporaba umetne inteligence se je začela razvijati tudi v poslovnem okolju. Analiza podatkov podjetjem in drugim institucijam omogoča lažje in učinkovitejše poslovanje. V nalogi oblikujemo model za napovedovanje denarnih prilivov podjetja. Model zasnujemo na preteklih računih izbranega podjetja, na katerih testiramo različne algoritme strojnega učenja. Izkaže se, da najboljše rezultate vrne algoritem XGBoost. Algoritem sodi med drevesne algoritme strojnega učenja in se uporablja tako za primere razvrščanja kot za regresijske primere modeliranja. Obravnavani model ocenimo z različnimi metodami za ocenjevanje modelov strojnega učenja. To so metrika natančnost, metrika AUC, ROC krivulja, krivulja preciznost - priklic, kalibracijska krivulja in druge. Model dodatno preizkusimo na novejših podatkih ter primerjamo rezultate ocen te napovedi z ocenami napovedi izvedene na testni množici podatkov.

• Language: Slovenian
• Keywords: napoved, denarni prilivi, XGBoost
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FMF - Faculty of Mathematics and Physics
• Year: 2022
• PID: 20.500.12556/RUL-139566
• COBISS.SI-ID: 120157955
• Publication date in RUL: 04.09.2022

OGNJANOVIĆ, Špela, 2022, Napovedovanje denarnih prilivov z algoritmom XGBoost : magistrsko delo [online]. Master’s thesis. [Accessed 5 April 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=139566

Secondary language

• Language: English
• Title: Cash inflow forecasting with XGBoost algorithm

Abstract

The use of artificial intelligence has also begun to develop in the business environment. Data analysis enables companies and other institutions to operate more easily and efficiently. In this work, we create a model for forecasting the company's cash inflows. We design the model on the past invoices of the selected company, on which we test various machine learning algorithms. It turns out that the XGBoost algorithm returns the best results. The algorithm belongs to the machine learning tree algorithms and is used for both classification and regression modeling examples. The considered model is evaluated with different methods for evaluating machine learning models. These are metric auccuracy, AUC metric, precision metric, ROC curve, precision-recall curve, calibration curve and others. We additionally test the model on more recent data and compare the results of the estimates of this forecast with the estimates of the forecast carried out on the test data set.

• Keywords: forecast, cash inflow, XGBoost