Supporting Proactive Process Recovery Decisions in
Agribusiness Using Business Intelligence Tools and Machine
Learning - Abstract
Athanasios Podaras 1
1
 Technical University of Liberec, Faculty of Economics, Department of Informatics, Studentská 1402/2, 46117,
Liberec 1, Czech Republic


                Summary
                Agricultural operations are continuously exposed to hazardous situations which can cause their
                unexpected disruptions. The experiences from the COVID-19 globally in the agrarian industry
                revealed the need for proactively defining policies to prevent the dramatic consequences of
                such unexpected business interruptions. A recent study from the Organization for Economic
                Co-operation and Development (OECD) (OECD, 2020) analyzes the necessity to rely on
                resilience – oriented policies to minimize the negative consequences of disruptive events in
                agriculture such as the pandemics or the natural hazards. Nevertheless, the ideal source for
                charting recovery policies are the continuity and process recovery data provided directly from
                the agribusiness practitioners. Nowadays, multiple companies which operate in the agrarian
                sector have started to chart continuity plans which can be activated in the event of a disruption.
                However, there is no evidence from the literature that these data are organized in sophisticated
                database management solutions. Due to the fact that big data is undoubtedly one of the assets
                that dominantly foster decisions nowadays, it can be realized that big data solutions for hosting,
                analyzing and manipulating continuity data in agribusiness is an issue of major importance.
                Business intelligence tools may support simultaneously decisions based on descriptive On-line
                Analytical Processing (OLAP) Reports as well as predictive rules induced via machine learning
                tools for more sophisticated data manipulation of the aforementioned data reports. To build
                such solutions, it is crucial to define data dimensions, namely Function, Agrarian Business
                Unit, Recovery Team, and Resumption Timeframe Required, as well as measurable facts such
                as estimated resumption timeframe, number of critical/non critical operations. A star,
                snowflake, star flake or constellation data warehouse schema can be utilized for the conceptual
                and the logical design. Moreover the SQL Server database management system may be utilized
                for the physical database solution. SQL Server may support subject-oriented and time-variant
                business intelligence solutions as well as OLAP reporting services. Machine learning
                predictive decisions can then be supported by classification or regression tools constructed
                using the exported data in combination with specialized software packages such as the R-
                Package for the induction of classifiers such as the decision trees, k-NN, k-Folds, SVM and
                naive bayes, as well as regression analysis patterns for the exported business continuity data.
                Additionally, specific data mining tasks, such as the association rule induction, can assist in
                exploring precise relationships among business continuity variables that affect the recovery
                procedure. The results which can stem from thorough business continuity big data processing
                and analysis can be used as feedback regarding risk management policy charting and process
                availability estimation in the agricultural industry.

                Keywords 1
                Agribusiness, business functions, business intelligence, continuity, data, decisions, machine
                learning, SQL, R-Package, recovery


Proceedings of HAICTA 2022, September 22–25, 2022, Athens, Greece
EMAIL: athanasios.podaras@tul.cz (A. 1)
ORCID: 0000-0003-4573-8976 (A. 1)
             ©️ 2022 Copyright for this paper by its authors.
             Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
             CEUR Workshop Proceedings (CEUR-WS.org)




                                                                                  415