Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Utforskande av en datadriven lösning för lagerhantering i små- och medelstora företag: En fallstudie av ett svenskt grossist- och detaljhandelsföretag
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2025 (Swedish)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
Exploring a Data-Driven Solution for Inventory Management in a Small to Medium-Sized Enterprise : A Case Study of a Swedish Wholesale and Retail Company (English)
Abstract [en]

This case study explores the performance of various machine learning (ML) models in forecasting demand using historical sales data from a Swedish small and medium-sized enterprise (SME). The dataset includes approximately 3200 unique articles over a time span of 19 months. After preprocessing, several models were trained and evaluated: Linear Regression, Decision Tree, Random Forest, and Gradient Boosting. Model performance was assessed using standard error metrics such as R2, RMSE and MAE. The results indicated that Gradient Boosting and Random Forest outperformed Linear Regression in terms of predictive accuracy, possibly with a marginal advantage under the constraints of limited data. Gradient Boosting provided highest accuracy at 77%, while Decision Tree only achieved an accuracy of 54%. The study underscores the potential of ML-based forecasting as a decision tool in inventory management, particularly within data-scarce environments typical of SMEs. However, considerations regarding implementation complexity, interpretability and cost benefit trade-offs suggest the need for caution and further investigations. These findings contribute to the understanding of ML applicability as a data-driven solution in SMEs.

Abstract [sv]

Denna fallstudie utforskar olika maskininlärningsmodellers (ML-modellers) förmåga att prognosticera efterfrågan, baserat på historiska försäljningsdata hämtat från ett svenskt små- och medelstort företag (SME). Datamängden innefattar cirka 3200 unika artiklar över en tidsperiod på 19 månader. Efter databearbetning, tränades och utvärderas flera modeller: Linjär regression, Decision tree, Random forest och Gradient boosting. Prestationen utvärderades med vanliga felmått som R2, RMSE, MAE. Resultaten indikerade att Gradient boosting och Random forest presterade bättre än Linjär regression i prediktiv noggrannhet, dock endast marginellt givet den begränsade datamängden. Gradient boosting uppvisade högst noggrannhet på 77%, medan Decision tree endast uppnådde en noggrannhet på 54%. Studien understryker ML-modellers prediktiva förmåga och potential som beslutsstöd vid lagerhantering, även i databegränsade miljöer som ofta kännetecknar SME. Dock tyder resultaten på utmaningar kring modellernas praktiska användbarhet i och med komplex implementation, tolkningsbarhet och kostnadsaspekter som bör tas i beaktande. Studien bidrar med insikter kring ML-modellers användbarhet som datadriven lösning i SMEs.

Place, publisher, year, edition, pages
2025. , p. 10
Series
TRITA-EECS-EX ; 2025:197
Keywords [en]
Decision Tree, Gradient Boosting, Industry 4.0, Inventory Management, Linear Regression, Machine learning (ML), Random Forest, Small and Medium-sized Enterprises (SMEs)
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-367600OAI: oai:DiVA.org:kth-367600DiVA, id: diva2:1985579
Supervisors
Examiners
Available from: 2025-07-29 Created: 2025-07-25 Last updated: 2025-07-29Bibliographically approved

Open Access in DiVA

fulltext(4041 kB)167 downloads
File information
File name FULLTEXT01.pdfFile size 4041 kBChecksum SHA-512
6884bc39726b568460f58db7eed38a39785624d3a4300ba7952c7a8621ea8bf3085393818b9fd3351dee952a7badb1ef792a0f801e209f416de0a686ba26176f
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 168 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 407 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf