A trend prediction model from very short term data learning

Currently, the environment has dynamic and changeable characteristics, making previously collected data unsuitable for building a predictive model, in that the value of sample population parameters such as mean or variance is moving or fluctuating. However, up-to-date data is usually in small sample sets, and it is risky to assume that the derived distribution; such as the normal distribution, from a few collected samples is an unbiased estimation of the underlying population. Based on this fact, the sample statistic [email protected]? may simply not be the proper measurement to estimate the mean of a population when confronting small data sets. This research proposes the Central Location Tracking Method (CLTM), with the novel concept of a ''trend center'', that is the center of probability (CP) determined by a variety of derived data properties which is employed to estimate the probable location of the population center @m. This approach aims at obtaining better predictability and fewer estimation errors for small sample sets. The comparison results between the method presented and [email protected]?, regression, neural networks, and ARIMA methods validate the superiority of this method for both random data and dependent data.

[1]  Claudio Moraga,et al.  A diffusion-neural-network for learning from small samples , 2004, Int. J. Approx. Reason..

[2]  Der-Chiang Li,et al.  Using mega-trend-diffusion and artificial samples in small data set learning for early flexible manufacturing system scheduling knowledge , 2007, Comput. Oper. Res..

[3]  Tomaso Poggio,et al.  Incorporating prior information in machine learning by creating virtual examples , 1998, Proc. IEEE.

[4]  F. G. Giesbrecht,et al.  Two-stage analysis based on a mixed model: large-sample asymptotic theory and small-sample simulation results , 1985 .

[5]  Der-Chiang Li,et al.  Using virtual sample generation to build up management knowledge in the early manufacturing stages , 2006, Eur. J. Oper. Res..

[6]  M. Kenward,et al.  Small sample inference for fixed effects from restricted maximum likelihood. , 1997, Biometrics.

[7]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[8]  G. Schwarz.  Estimating the Dimension of a Model , 1978 .

[9]  J. Ware,et al.  Random-effects models for longitudinal data. , 1982, Biometrics.

[10]  D. Harville.  Maximum Likelihood Approaches to Variance Component Estimation and to Related Problems , 1977 .

[11]  M. Schluchter,et al.  Small-sample adjustments to tests with unbalanced repeated measures assuming several covariance structures , 1990 .

[12]  R. Jennrich,et al.  Unbalanced repeated-measures models with structured covariance matrices. , 1986, Biometrics.

[13]  Chun-Wu Yeh,et al.  A non-parametric learning algorithm for small manufacturing data sets , 2008, Expert Syst. Appl..