The M4 competition: Bigger. Stronger. Better

Petropoulos, Fotios; Makridakis, Spyros

doi:10.1016/j.ijforecast.2019.05.005

Cited by 14 publications

(4 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to train the CustomGPT, we fed into it all the papers from the M3 [4], M4 [5], and M5 [6] special issues published by the International Journal of Forecasting as well as papers from the ScienceDirect database by searching "M competition OR M2 competition OR M3 competition OR M4 competition OR M5 competition" in the Title field and "forecast" in the "Title, abstract, keywords" field. Note that in the ScienceDirect search API (https: //dev.elsevier.com/sd_apis.html, accessed on 18 August 2023), punctuation is ignored in a phrase search, so the searches "M3 competition" and "M3-competition" return the same results.…”

Section: The Accuracy Of the Standard And Customized Chatgptmentioning

confidence: 99%

Large Language Models: Their Success and Impact

Makridakis,

Petropoulos,

Kang

2023

Forecasting

Self Cite

View full text Add to dashboard Cite

ChatGPT, a state-of-the-art large language model (LLM), is revolutionizing the AI field by exhibiting humanlike skills in a range of tasks that include understanding and answering natural language questions, translating languages, writing code, passing professional exams, and even composing poetry, among its other abilities. ChatGPT has gained an immense popularity since its launch, amassing 100 million active monthly users in just two months, thereby establishing itself as the fastest-growing consumer application to date. This paper discusses the reasons for its success as well as the future prospects of similar large language models (LLMs), with an emphasis on their potential impact on forecasting, a specialized and domain-specific field. This is achieved by first comparing the correctness of the answers of the standard ChatGPT and a custom one, trained using published papers from a subfield of forecasting where the answers to the questions asked are known, allowing us to determine their correctness compared to those of the two ChatGPT versions. Then, we also compare the responses of the two versions on how judgmental adjustments to the statistical/ML forecasts should be applied by firms to improve their accuracy. The paper concludes by considering the future of LLMs and their impact on all aspects of our life and work, as well as on the field of forecasting specifically. Finally, the conclusion section is generated by ChatGPT, which was provided with a condensed version of this paper and asked to write a four-paragraph conclusion.

show abstract

Section: The Accuracy Of the Standard And Customized Chatgptmentioning

confidence: 99%

Large Language Models: Their Success and Impact

Makridakis,

Petropoulos,

Kang

2023

Forecasting

Self Cite

View full text Add to dashboard Cite

show abstract

“…Such methods include simple extrapolative methods, such as exponential smoothing and moving averages (Gardner Jr 1985, Svetunkov and, autoregressive integrated moving average (ARIMA)-type models (Gilbert 2005, machine learning methods (Zhang andQi 2005, Punia et al 2020), and judgmental methods (Petropoulos et al 2016(Petropoulos et al , 2018. A considerable amount of research work has been dedicated to analyzing and comparing the performance of such methods through empirical investigations and international forecasting competitions using supply chain data (Petropoulos and Makridakis 2020). The last forecasting competition (referred to as the M5 competition) was built on the case of a Walmart retail supply chain with more than 30,000 products (Makridakis et al 2020).…”

Section: Demand Forecasting In the Retail Contextmentioning

confidence: 99%

Basket data-driven approach for omnichannel demand forecasting

Klibi

Babaï

Ducq

et al. 2023

International Journal of Production Economics

View full text Add to dashboard Cite

“…The training data set consists of 3049 different items in ten different Walmart stores located in three different US states, thereby comprising 30,490 time series in total. Given the important role of the M competitions in the development of new forecasting methods ( Petropoulos & Makridakis, 2020 ), we believe that this data set is an important benchmark case not only for forecasting, but also for intermittent demand inventory control. In this section we briefly recap the data characteristics and then discuss how we assess the goodness-of-fit of memoryless demand processes.…”

Section: Data and Goodness-of-fit Of Memoryless Demand Interval Modelsmentioning

confidence: 99%