دانلود رایگان مقاله لاتین عملکرد الگوریتم پیش بینی از سایت الزویر


عنوان فارسی مقاله:

تجسم عملکرد الگوریتم پیش بینی با استفاده از سری زمانی به عنوان مثال فضاها 


عنوان انگلیسی مقاله:

Visualising forecasting algorithm performance using time series instance spaces


سال انتشار : 2017



برای دانلود رایگان مقاله عملکرد الگوریتم پیش بینی اینجا کلیک نمایید.





بخشی از مقاله انگلیسی:


4. Comparison of time series forecasting methods in the instance space

 The No-Free-Lunch theorem was proposed for supervised machine learning by Wolpert (1996) and for search and optimisation by Wolpert and Macready (1997). It tells us that there is never likely to be a single method that fits all situations. Similarly, there is no one time series forecasting method that will always perform best. Even for one particular time series, no one technique is consistently superior to all others (Lawrence, 2001). Petropoulos, Makridakis, Assimakopoulos, and Nikolopoulos (2014) wrote, ‘‘as there are ‘horses for courses’, there must also be forecasting methods that are more tailored to some types of data’’, and measured the extent of the effects of seven time series features on the forecasting accuracy. Smith-Miles et al. (2014) proposed a method for comparing and visualising the strengths and weaknesses of different graph colouring algorithms across an instance space. Here, we consider six general time series forecasting methods in order to demonstrate the potential of the instance space for algorithm performance visualisation. These are:abstract

It is common practice to evaluate the strength of forecasting methods using collections of well-studied time series datasets, such as the M3 data. The question is, though, how diverse and challenging are these time series, and do they enable us to study the unique strengths and weaknesses of different forecasting methods? This paper proposes a visualisation method for collections of time series that enables a time series to be represented as a point in a two-dimensional instance space. The effectiveness of different forecasting methods across this space is easy to visualise, and the diversity of the time series in an existing collection can be assessed. Noting that the diversity of the M3 dataset has been questioned, this paper also proposes a method for generating new time series with controllable characteristics in order to fill in and spread out the instance space, making our generalisations of forecasting method performances as robust as possible. © 2016 International Institute of Forecasters. Published by Elsevier B.V. All rights reserved.Naïve: using the most recent observation as the forecast for all future periods. • Seasonal naïve: forecasts are equal to the most recent observation from the corresponding time of year. For yearly data, this is equivalent to the naïve method. • The Theta method, which performed particularly well in the M3-Competition (Assimakopoulos & Nikolopoulos, 2000; Makridakis & Hibon, 2000). We apply the method as implemented in the stheta function from the forecTheta package (Fiorucci, Louzada, & Yiqi, 2016). • ETS: exponential smoothing state space modelling (Hyndman, Koehler, Snyder, & Grose, 2002), which is used widely as a general forecasting algorithm for trended and seasonal time series. • ARIMA: autoregressive integrated moving average models, as implemented rin the automated algorithm of Hyndman and Khandakar (2008). • STL-AR: an AR model is fitted to the seasonally adjusted series obtained from a STL decomposition (Hyndman & Athanasopoulos, 2014), while the seasonal component is forecast using the seasonal naïve method. The two forecasts are summed to obtain forecasts of the original time series.



برای دانلود رایگان مقاله عملکرد الگوریتم پیش بینی اینجا کلیک نمایید.






کلمات کلیدی:

ECAI 2016: 22nd European Conference on Artificial Intelligence, 29 ... https://books.google.com/books?isbn=1614996725 G.A. Kaminka, ‎M. Fox, ‎P. Bouquet - 2016 - ‎Computers Online. Prediction. of. Exponential. Decay. Time. Series. with. Human-Agent ... which theoretically compares the performance of our online algorithm relative to the performance ... In each round, the learner first receives a time series instance. Business Information Systems: 15th International Conference, BIS ... https://books.google.com/books?isbn=3642303595 Witold Abramowicz, ‎Dalia Kriksciuniene, ‎Virgilijus Sakalauskas - 2012 - ‎Business & Economics Let ON(I) be the return of algorithm ON on input instance I ∈ I . Let OPT be the ... features in time series with the objective to maximize over all performance. Advances in Information Technology and Education: International ... https://books.google.com/books?isbn=3642224172 Honghua Tan, ‎Mark Zhou - 2011 - ‎Computers Therefore, among the many nonlinear time series models developed include bilinear, ... For instance, prior signal processing knowledge is required to adjust specific ... Unfortunately, the standard BPNN using the steepest descent algorithm has the ... Performance of the proposed SOPNN scheme is evaluated by using a ... Data Classification: Algorithms and Applications https://books.google.com/books?isbn=1466586745 Charu C. Aggarwal - 2014 - ‎Business & Economics Huerta et al. extend SVMs to classify multidimensional time series by using ... et al. propose a method that improves SVM time series classification performance by ... algorithm to transform the time series, by splitting each time series instance ... Knowledge and Systems Engineering: Proceedings of the Sixth ... https://books.google.com/books?isbn=3319116800 Viet-Ha Nguyen, ‎Anh-Cuong Le, ‎Van-Nam Huynh - 2014 - ‎Computers In each subcluster Si, we select randomly a time series from this subcluster to be its ... If that is the case, we check if each instance in S j is similar to C_Si using the ... We evaluate the performance of our proposed algorithms for discovering ...