Viðburðir eftir árum


Meistaravörn við tækni- og verkfræðideild

Sales forecasting using different forecasting methods

  • 7.6.2019, 9:00 - 10:00

On Friday June 7th at 9:00, Marta Rut Ólafsdóttir will defend her 30 ECTS thesis in MSc Engineering Management. The title of the thesis is "Sales forecasting using different forecasting methods" and the lecture will be held in room M209. All welcome.

Student: Marta Rut Ólafsdóttir
Supervisor: Eyjólfur Ingi Ásgeirsson
Examiner: Michal Borsky

Title: Sales forecasting using different forecasting methods

Abstract:
The advantages to accurately forecasting sales are significant. For any company it is important to have foresight knowledge of financial outcomes and to have confidence in the forecasting process to be able to trust its results. This knowledge is the basis for all operations planning and makes the company better equipped to deal with situations that may arise. Recently machine learning has become the new buzzword in business and business leaders are keen to find out if these methods are applicable in today’s business environment. One of the main challenges for businesses, when striving to adapt machine learning in their processes, is lack of appropriate data. Performance of machine learning models has been associated with access to large amounts of data that enable the models to learn. This thesis examines if published external data can be used to generate accurate sales forecasts for the medical devices company Össur.

Two different approaches were used; traditional time series methods using only historical quarterly sales data from 2001 to 2018 and machine learning methods using the historical sales data as well as exogenous variables believed to influence sales. The traditional time series models that were applied were simple moving average, decomposition using least squares regression and Holt-Winters. The machine learning models applied were multiple linear regression, random forest and neural networks. The accuracy of the models was then compared using the RMSE value for the testing data set. The machine learning methods all yielded lower RMSE values then the traditional time series methods. The model the yielded the lowest RMSE value was the random forest model.



Vinsamlegast athugið að á viðburðum Háskólans í Reykjavík (HR) eru teknar ljósmyndir og myndbönd sem notuð eru í markaðsstarfi HR. Hægt er að nálgast frekari upplýsingar á ru.is eða með því að senda tölvupóst á netfangið: personuvernd@ru.is
//
Please note that at events hosted at Reykjavik University (RU), photographs and videos are taken which might be used for RU marketing purposes. Read more about this on our ru.is or send an e-mail: personuvernd@ru.is