A Comparison of Time Series Forecasting Methods

produce a systematic Literature Review according to the rules of D. Denyer and D. Tranfield (2009) “Chapter 39: producing a systematic review”. I'm doing an Msc. in Transport and Supply Chain Management (Business Administration) and my topic focusses on Forecasting Methods/Models, more specifically Time Series models. As attachments you can find the following documents: a) Guidelines LR - Thesis This includes the guidelines that the University has given me. b) Literature review - Thesis This concerns my first draft that I've already handed in previously. However, this draft has not been approved as it is stated to be "Too Easy". More feedback on this draft can be find here below, which can be used as direction points. My thesis will include the following Main RQ: Which extrapolative time series model is most accurate in forecasting the future demand of "Company X"? Please note that this RQ has been approved and in my thesis I will built 6 different forecasting models that I compare to each other and eventually be able to recommend the company a certain forecasting method. Currently, the Naive method, Simple Exponential smoothing method, Holts-winter and arima can be included. However, the university believes that these models are too easy. I need to include 1 more model that is more complex and spice-up the whole review. Detailed feedback from the first draft: - No explanation why you are using time series models; give reasons why this is what you chose - Please use the "producing a systematic review" article from Denyer. - wrong referencing, wrong usement of abbreviations - Overall review is too simple and sounds too "lecturing". There is no real discussion or gap analysis. - Provide conceptual framework, use tables where you verify with which forecasting methods other researchers used and what their outcomes were - No figures inside the text, make your own tables - Too many sections, don't name each paragraph! - Section 2.3 is more a methodology part; don't 'teach', you should discuss. In which cases/situation would certain forecasting measures be more convenient than others? Which measures do you pick and which ones don't you pick? What would you do when demand is frequently 0? Which accuracy measure is recommended by authours? - Only at the end of the LR you finish with the forecasting methods that you will pick for your thesis, but in the core content you should discuss much more methods. Don't limit it already to the ones that you will continue with. - "Company X" currently only uses Judgemental Forecasting, but no method that is based on historical data. Try to outline this in the LR as well. Please read the first draft to understand the direction I'd like to go. However, there is no need to 'rewrite' this version, it is more than ok to start      

Sample Solution