Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa
Thesis (MCom)--Stellenbosch University, 2022.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2022
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614079101698048 |
|---|---|
| access_status_str | Open Access |
| author | Gouws, Jauné |
| author2 | Lubbe, Sugnet |
| author_browse | Gouws, Jauné Lubbe, Sugnet |
| author_facet | Lubbe, Sugnet Gouws, Jauné |
| author_sort | Gouws, Jauné |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MCom)--Stellenbosch University, 2022. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/126089 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:46:20.037Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2022 |
| publishDateRange | 2022 |
| publishDateSort | 2022 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/126089 Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa Gouws, Jauné Lubbe, Sugnet Stellenbosch University. Faculty of Economic and Management Sciences. Dept. of Statistics and Actuarial Science. Discriminant analysis Kernel functions Mathematical statistics UCTD Thesis (MCom)--Stellenbosch University, 2022. ENGLISH SUMMARY: The objective of this thesis is to provide an overview of two sets of classification approaches to a real-world problem found in corporate valuation practice in South Africa. The first being a probabilistic approach, which includes algorithms such as Fisher’s Linear Discriminant Analysis (FLDA), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA) and Kernel Fisher Discriminant Analysis (KFDA). The second approach, being a deterministic approach, includes techniques such as the Perceptron, the Support Vector Classifier (SVC) and the Support Vector Machine (SVM). The above listed techniques will be introduced by means of a literature study, both in the two-class and multi-class case. However, emphasis will be placed on the two-class case. As an overview, classification is a widely used statistical procedure to classify objects into two or more classes according to some rule, based on the input variables. However, classification of objects with these methods can get complicated when classes are not linearly separable. This problem introduces another layer to the objective of this thesis, which is to consider the viability of applying the above listed techniques in both a linear and non-linear setting. We will see that techniques utilised in a non-linear setting, are often derived by generalising the algorithm with a kernel function. Following the introduction of the various classification techniques, a detailed introduction will be provided on the real-world problem that corporate valuation practitioners in South Africa face. Very little empirical guidance is available regarding best valuation practices in emerging markets, such as South Africa. As such, Bruner, Conroy, Estrada, Kritzman and Li (2002:311) urged for the development of a best practice guide for analysts in emerging markets. The specific contribution that the application of the respective classification algorithms on this problem aims to make, is in the field of relative valuations, otherwise known as multiples. AFRIKAANSE OPSOMMING: Die doelwit van hierdie tesis is om 'n oorsig te gee van twee stelle klassifikasiebenaderings tot 'n werklike probleem wat in korporatiewe waardasiepraktyke in Suid-Afrika voorkom. Die eerste is 'n waarskynlikheidsbenadering, wat algoritmes soos Fisher se Lineere Diskriminant Analise (FLDA), Lineere Diskriminant Analise (LDA), Kwadratiese Diskriminant Analise (KDA) en Kern Fisher Diskriminant Ananalise (KFDA) insluit. Die tweede benadering, synde 'n deterministiese benadering, sluit tegnieke soos die Perceptron, die Steun Vektor Klassifiseerder (SVK) en die Steun Vektor Masjien (SVM) in. Die bogenoemde tegnieke sal deur middel van 'n literatuurstudie bekendgestel word, beide in die twee-klas en multi-klas geval. Klem sal egter op die tweeklassaak gele word. As 'n oorsig is klassifikasie 'n algemeen gebruikte statistiese prosedure om objekte in twee of meer klasse te klassifiseer volgens een of ander reel wat op die onafhanklike veranderlikes gebaseer is. Klassifikasie van objekte met hierdie metodes kan egter ingewikkeld raak wanneer klasse nie lineer skeibaar is nie. Hierdie probleem stel nog 'n lens in tot die doelwit van hierdie tesis, wat is om die lewensvatbaarheid van die toepassing van die bogenoemde tegnieke in beide 'n lineere en nie-lineere omgewing te oorweeg. Ons sal sien dat tegnieke wat in 'n nie-lineere omgewing gebruik word, dikwels afgelei word deur die algoritme te veralgemeen deur die gebruik van 'n kernfunksie. Na die bekendstelling van die verskillende klassifikasietegnieke, sal 'n gedetailleerde inleiding verskaf word oor die werklike probleem waarmee korporatiewe waardasiepraktisyns in Suid-Afrika te kampe het. Baie min empiriese navorsing is beskikbaar rakende beste waardasiepraktyke in ontluikende markte, soos Suid-Afrika. As sodanig het Bruner, Conroy, Estrada, Kritzman en Li (2002:311) aangedring op die ontwikkeling van 'n beste praktykgids vir ontleders in ontluikende markte. Die spesifieke bydrae wat die toepassing van die onderskeie klassifikasie-algoritmes op hierdie probleem beoog om te lewer, is op die gebied van relatiewe waardasies, ook bekend as veelvoude. Masters 2022-11-21T11:17:06Z 2023-01-16T12:49:32Z 2022-11-21T11:17:06Z 2023-01-16T12:49:32Z 2022-12 Thesis http://hdl.handle.net/10019.1/126089 en_ZA Stellenbosch University xvii, 118 pages : illustrations, includes annexures application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Discriminant analysis Kernel functions Mathematical statistics UCTD Gouws, Jauné Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title | Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title_full | Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title_fullStr | Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title_full_unstemmed | Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title_short | Comparison of classification techniques to review the plausibility of adjustments to multiples-based equity valuations : empirical study for South Africa |
| title_sort | comparison of classification techniques to review the plausibility of adjustments to multiples based equity valuations empirical study for south africa |
| topic | Discriminant analysis Kernel functions Mathematical statistics UCTD |
| url | http://hdl.handle.net/10019.1/126089 |
| work_keys_str_mv | AT gouwsjaune comparisonofclassificationtechniquestoreviewtheplausibilityofadjustmentstomultiplesbasedequityvaluationsempiricalstudyforsouthafrica |