Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods
Thesis (MEng)--Stellenbosch University, 2023.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2023
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614113500233728 |
|---|---|
| access_status_str | Open Access |
| author | Bester, Morne |
| author2 | Van Eeden, Joubert |
| author_browse | Bester, Morne Van Eeden, Joubert |
| author_facet | Van Eeden, Joubert Bester, Morne |
| author_sort | Bester, Morne |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2023. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/127041 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA en_ZA |
| last_indexed | 2026-06-10T12:46:52.985Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2023 |
| publishDateRange | 2023 |
| publishDateSort | 2023 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/127041 Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods Bester, Morne Van Eeden, Joubert Van Rooyen, Gert-Jan Stellenbosch University. Faculty of Engineering. Dept. of Industrial Engineering. Citrus fruit industry -- Technological innovations Agricultural innovations Harvesting -- Computer programs Economic forecasting Machine learning Neural networks (Computer science) Thesis (MEng)--Stellenbosch University, 2023. ENGLISH ABSTRACT: Predicting the supply of fresh fruit is critical for various stakeholders throughout the supply chain. Growers require supply predictions to plan for harvesting activities and apply the appropriate production practices. Packing facilities also have to plan their capacity and inventory accordingly. On the other hand, exporters require accurate supply predictions to satisfy client demand and establish an optimal market plan. These decisions and planning activities are also critical for the citrus industry and its relevant stakeholders. Many decisions are based on the supply predictions of citrus and having an inaccurate prediction can lead to several losses, including opportunity losses, dissatisfied clients and unplanned or unnecessary expenses. Improved supply prediction performance can combat these losses. However, improving the citrus supply prediction can be difficult for an individual or stakeholder. Furthermore, many different techniques and technologies are available to address specific components or features of the supply prediction problem. Furthermore, there is likely no universal superior supply prediction method. Stakeholders may have different evaluation criteria, other than prediction performance, to classify if a method is suitable for the problem at hand. For example, cost, duration, and complexity are some stakeholder-specific evaluation criteria to consider. Consequently, a toolkit was developed to address the issue at hand—the toolkit comprised of two main parts, i.e. a toolset and activities. The toolset comprises six categories, including a repository of techniques used to synthesise and evaluate a supply prediction. Synthesising a method is based on sampling designs, establishing a sample size, selecting data collection technologies, and using prediction or estimation techniques. Evaluation techniques, however, include performance metrics and general evaluation criteria such as cost, duration, and limitation analysis. On the other hand, the toolkit activities consist of several guides to assist an individual with evaluating and improving a citrus supply prediction. These evaluation guides include method performance, cost, duration, and limitation guides. The toolkit was applied in a case study where a scenario analysis identified the size distribution prediction as the main improvement area. Consequently, several methods were synthesised to improve the size distribution prediction and, subsequently, the supply prediction performance. The alternative methods were synthesised by following the method improvement principles and considering different techniques from the toolset. Consequently, the alternative methods were trialled and evaluated in five orchards. An appropriate alternative method was identified, which improved the supply prediction performance for the five orchards. This method can be implemented in the following season(s). AFRIKAANS OPSOMMING: Die voorspelling van vars vrugte verskaffing is krities vir verskeie belanghebbendes regdeur die voorsieningsketting. Produsente benodig voorsiening voorspellings om vir oes verwante aktiwiteite te beplan asook om toepaslike produksie aktiwiteite toe te pas. Pak-fasiliteite moet ook hul kapasiteit en voorraad beplanning kan doen. Aan die ander kant benoding uitvoerders akkurate voorsiening voorspellings om kli¨ente vraag te bevredig asook om ’n optimale mark-plan te vestig. Hierdie besluite en beplannings aktiwiteite het ook ’n kritiese invloed op die gehele sitrus industrie en sy relevante belanghebbendes. Vele besluite is gebaseer op die voorsiening voorspelling van sitrus en onakkurate voorspellings kan lei tot verskeie verliese, insluitend geleetheidsverliese, onbevredigde kli¨ente asook onbeplande en onnodige uitgawes. ’n Verbeterde voorsiening voorspelling kan hierdie verliese teenstand bied. Dit kan egter uitdagend wees vir ’n individu of belanghebbende om hul voorsiening voorspelling te verbeter. Verder is daar ook verskeie tegnieke en tegnologie¨e beskikbaar vir komponente van ’n voorsiening voorspelling om sodoende die algehele voorsiening voorspelling probleem aan te spreek. Daar is ook waarskynlik geen universele beste voorsiening voorspelling metode nie. Belanghebbendes mag verskeie evaluasie kriteria hˆe, anders as voorspelling uitnemendheid, om te klassifiseer of ’n metode geskik is vir die probleem. Byvoorbeeld, koste, tydsduur en kompleksiteit is sekere belanghebbende-spesifieke evaluasie kriteria om te oorweeg. Gevolglik, was ’n sogenaamde gereedskapstel ontwikkel om die uitdaging aan te spreek – die gereedskapstel bestaan uit twee hoofdele, naamlik die gereedskap en aktiwiteite. Die gereedskap kan in ses kategorie¨e geklassifiseer word, insluitend ’n bundel tegnieke wat gebruik kan word om voorsiening voorspelling metodes te vestig (of sintetiseer) en te evalueer. Om ’n metode te vestig is gebaseer op die monsterneming ontwerp, die grootte van die monster, keuse van data inname tegnologie asook die gebruik van voorspelling tegnieke. Evaluasie tegnieke sluit in verskeie prestasie maatstawwe en algemene evaluasie kriteria soos koste, tydsduur en beperkings analises. Die gereedskapstel aktiwiteite sluit in voorgestelde riglyne om die individu of belanghebbende te ondersteun met hul evalueering en verbetering van ’n sitrus voorsiening voorspelling. Die evaluasie riglyne behels dus om die metode se prestasie, koste, tydsduur en beperkinge te analiseer en evalueer. Die gereedskapstel was toegepas in ’n gevallestudie waar, m.b.v. scenario analise, die grootte verspreiding voorspelling ge¨ıdentifiseer was as die hoof verbeterings area. Gevolglik was verskeie metodes gevestig om die grootte verspreiding voorspelling te verbeter en sodoende die voorsiening voorspelling te verbeter. Die alternatiewe metodes was bepaal d.m.v. gebruik te maak van die verbetering beginsels riglyne asook die tegnieke van die gereedskapstel. Die alternatiewe metodes was beproef en ge¨evalueer in vyf boorde. ’n Toepaslike alternatiewe metode was ge¨ıdentifiseer wat die voorsiening voorspelling verbeter het vir die vyf boorde. Hierdie metode sal in die volgende seisoen(e) toegepas kan word. Masters 2023-01-31T19:28:58Z 2023-05-18T07:01:26Z 2023-01-31T19:28:58Z 2023-05-18T07:01:26Z 2023-01 Thesis http://hdl.handle.net/10019.1/127041 en_ZA en_ZA Stellenbosch University xxiv, 180 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Citrus fruit industry -- Technological innovations Agricultural innovations Harvesting -- Computer programs Economic forecasting Machine learning Neural networks (Computer science) Bester, Morne Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title | Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title_full | Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title_fullStr | Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title_full_unstemmed | Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title_short | Development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| title_sort | development of a toolkit for the evaluation and improvement of commercial citrus supply prediction methods |
| topic | Citrus fruit industry -- Technological innovations Agricultural innovations Harvesting -- Computer programs Economic forecasting Machine learning Neural networks (Computer science) |
| url | http://hdl.handle.net/10019.1/127041 |
| work_keys_str_mv | AT bestermorne developmentofatoolkitfortheevaluationandimprovementofcommercialcitrussupplypredictionmethods |