1 of 18

X International conference�“Information Technology and Implementation” (IT&I-2023)�Kyiv, Ukraine

Vadym Pakholchuk

Ph. D. in Finance

Tetiana Zatonatska

doctor of economic sciences, professor

Alim Syzov

Candidate of economic science, associate professor

Daryna Vorontsova

student

1

Dedicated to the tenth anniversary of the Faculty of Information Technology

Implementation of Data Science Methods in Armed Forces Budgeting: Challenges and Opportunities

Annotation. This paper explores the challenges and opportunities of incorporating data science methods into military budgeting, with a focus on practical advice in the area of forecasting defense budgets and expenditures. The study uses aggregated monthly expenditure data from military bases and applies various forecasting models, including smoothing, linear regression, prophet and XGBoost. The three best performing models are selected to build a meta-model that improves the accuracy of the forecasts. The results suggest that incorporating data science methods into military budgeting can lead to better decision making and resource optimization

Keywords: Data Science, armed forces budgeting, forecasting, defense budget, defense expenditures .

In the process of creating of information technologies, mathematical models and methods are widely used, among them those that relate to multicriteria (vector) optimization. Let us consider the vector optimization problem under the condition of possible disturbances in the input data of the vector criterion consisting of quadratic functions , where ‒ set of arbitrary structure, probably discrete, , is an n-dimensional real space, , , , , , , ‒ space of problem input data relating to the vector criterion.

For any input data and sufficiently small number let us define the set of perturbed input data and the perturbed problem .

We will consider the problem in two versions: 1) ‒ to find elements of the Pareto set , where ; 2) ‒ to find elements of the set of Slater optimal solutions where .

2 of 18

IMPLEMENTATION OF DATA SCIENCE METHODS IN ARMED FORCES BUDGETING: CHALLENGES AND OPPORTUNITIES�

Transition to NATO standards in the area of financing the Armed Forces;
rapid response to the potential needs of the Armed Forces;
building forecast scenarios based on a large data sets and adjusting to them;
transparency of the use of budgetary funds through full automatetion of the budgeting and management process using DS tools;

Information Technology and Implementation, November 20, 2023, Taras Shevchenko National University of Kyiv, Kyiv, Ukraine

3 of 18

Data science in modern budgeting

Domain

Computer Science

Math

Public Finance

Defense Finance

4 of 18

What areas of Data Science should be mentioned?

Clustering

Sentiment

Forecasting

Classification

NLP

5 of 18

Data collection and analysis

6 of 18

Cascade modeling

Base model assumption

SMA

Exponential smoothing (Holt-Winter)
Rolling mean

Times Series Models

SARIMA
Prophet

ML Models

Liner Regression
XGBoost

7 of 18

Simple Rolling mean forecast

8 of 18

Exponential smoothing

9 of 18

Double exponential smoothing

10 of 18

Triple exponential smoothing

11 of 18

Classic TS SARIMA models

ARIMA (p, d, q) x (P, D, Q) S =

ARIMA (0, 0, 0) x (3, 1, 2)12

12 of 18

Meta Prophet as a SOTA model in time series forecasting

This model includes g(t) - a non-linear trend model that allows for considering trends in the context of forecast horizons, s(t) - seasonality, and h(t) - holidays and important events, as well as an error term e

13 of 18

Good old linear regression

14 of 18

Boosting is an old new fashion

15 of 18

Explainability always matter

SHAP-values for linear regression

SHAP-values for boosting

16 of 18

Building the best mix of models

Metrics	Holt-Winter	SARIMA	Prophet	Regression	XGBoost
MAE	606972	1682813	812210	595491	567103
MAPE	8.42%	24.50%	12.49%	7.81%	7.16%

17 of 18

The Problem with High-Scale Forecasting

Source: https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create-time-series

18 of 18

Final thoughts and further �research ideas

Data science methods can enable the Armed Forces to make more accurate and comprehensive forecasts
Become increasingly critical for effective budgeting and decision-making in the Armed Forces
Improve mission readiness, resource allocation, and cost-effectiveness
Investing in data collection and management systems and personnel training