Carton and volume forecasting from picking lines
dc.contributor.advisor | Visagie, Stephan E. | en_ZA |
dc.contributor.author | Samuels, Jason | en_ZA |
dc.contributor.other | Stellenbosch University. Faculty of Economic and Management Sciences. Dept. of Logistics. Logistics. | en_ZA |
dc.date.accessioned | 2017-02-09T17:22:11Z | |
dc.date.accessioned | 2017-03-29T11:32:14Z | |
dc.date.available | 2017-02-09T17:22:11Z | |
dc.date.available | 2017-03-29T11:32:14Z | |
dc.date.issued | 2017-03 | |
dc.description | Thesis (MCom)--Stellenbosch University, 2017. | |
dc.description.abstract | ENGLISH SUMMARY : Supply chains consist of many stages and all these stages need to be managed. Being able to predict stock flow at any stage can be cost effective for the whole supply chain. In this thesis data from the Pepstores Ltd (PEP) distribution centre in Kuilsrivier, Cape Town are used to predict number of cartons and volume of stock that a hub in their supply chain owned by Pepkor logistics (PKL) will receive. These forecasts will help PKL to schedule delivery trucks and routes to stores with more accurate data and thus lower transportation costs. Simple linear regression (SLR) and multiple linear regression (MLR) are used to predict cartons and volume, but heteroscedasticity is obtained in the residuals. Different types of transformations on the SLR model are introduced and used on dependent and independent variables. A logarithmic weighted transformation could overcome these problems and is thus used along with polynomial regression to predict the number of cartons and volume of stock. The carton prediction model uses a polynomial regression model with order 2 and the volume prediction model uses a SLR model on the logarithmic weighted variables. Accuracy tests show that the models predict the number of cartons and volumes of stock well. A case study on actual data to forecast volume and cartons is presented. These predictions were then compared to the actual values and the forecast that was sent to the hub from the DC over a two week period. It is concluded that PEP can use these models within their systems, but coecients need to be reviewed periodically in order to take into account the different types of products. | en_ZA |
dc.description.abstract | AFRIKAANSE OPSOMMING : Voorsieningskettings bestaan uit verskeie stappe en elke stap moet bestuur word. Die vermoe om voorraadvloei te kan bepaal by enige stap kan voordelig wees vir die hele voorsieningsketting. In hierdie tesis word data van Pepstores Bpk (PEP) se verspreidingsentrum in Kuilsrivier, Kaapstad gebruik om die aantal kartonne en die volume voorraad te voorspel wat na 'n Pepkor Logistics (PKL) verspreidingsentrum gestuur word. Hierdie voorspellings sal vir PKL help om roetes na takke met akkurater data te bepaal en dus kostes te bespaar. Eenvoudige linieere regressie en meervoudige linieere regressie word gebruik om die kartonne en volume te bepaal, maar heteroskedastisiteit kom voor in die residue. Verskillende tipe transformasies word voorgestel op die eenvoudige linieere regressiemodel en word gebruik op die afhanklike en onafhanklike veranderlikes. 'n Logaritmiese geweegde transformasie saam met polinomiese regressie word gebruik om die aantal kartonne en volume voorraad te voorspel. Die kartonvoorspellingsmodel gebruik 'n polinomiese regressie van orde 2 terwyl die volumevoorspellingsmodel 'n eenvoudige linieere regressiemodel gebruik op die logaritmiese geweegde veranderlikes gebruik. Die akkuraatheid van vooruitskattings word getoets en die modelle voorspel die aantal kartonne en volume goed. 'n Gevallestudie met werklike data wat volumes en kartonne voorspel word aangebied. Die voorspellings is dan vergelyk met die werklike waardes en die voorspellings wat gestuur was van die PEP verspreidingsentrum na die sekondere verspreidingsentrum. Daar word bevind dat PEP hierdie modelle kan gebruik in hulle stelsels, maar dat die koesiente van die model moet gereeld hersien word om die veranderende verkoopspatrone van produkte in ag te neem. | af_ZA |
dc.format.extent | viii, 128 pages ; illustrations | |
dc.identifier.uri | http://hdl.handle.net/10019.1/100797 | |
dc.language.iso | en_ZA | en_ZA |
dc.publisher | Stellenbosch : Stellenbosch University | |
dc.rights.holder | Stellenbosch University | |
dc.subject | Forecasting -- Statistical methods | en_ZA |
dc.subject | Regression analysis | en_ZA |
dc.subject | Warehouses -- Kuils River (Western Cape, South Africa) | en_ZA |
dc.subject | UCTD | |
dc.title | Carton and volume forecasting from picking lines | en_ZA |
dc.type | Thesis | en_ZA |