A Run Length Transformation for Discriminating Between Auto Regressive Time Series

Bagnall, Anthony; Janacek, Gareth
July 2014
Journal of Classification;Jul2014, Vol. 31 Issue 2, p154
Academic Journal
We describe a simple time series transformation to detect differences in series that can be accurately modelled as stationary autoregressive (AR) processes. The transformation involves forming the histogram of above and below the mean run lengths. The run length (RL) transformation has the benefits of being very fast, compact and updatable for new data in constant time. Furthermore, it can be generated directly from data that has already been highly compressed. We first establish the theoretical asymptotic relationship between run length distributions and AR models through consideration of the zero crossing probability and the distribution of runs. We benchmark our transformation against two alternatives: the truncated Autocorrelation function (ACF) transform and the AR transformation, which involves the standard method of fitting the partial autocorrelation coefficients with the Durbin-Levinson recursions and using the Akaike Information Criterion stopping procedure. Whilst optimal in the idealized scenario, representing the data in these ways is time consuming and the representation cannot be updated online for new data. We show that for classification problems the accuracy obtained through using the run length distribution tends towards that obtained from using the full fitted models. We then propose three alternative distance measures for run length distributions based on Gower's general similarity coefficient, the likelihood ratio and dynamic time warping (DTW). Through simulated classification experiments we show that a nearest neighbour distance based on DTW converges to the optimal faster than classifiers based on Euclidean distance, Gower's coefficient and the likelihood ratio. We experiment with a variety of classifiers and demonstrate that although the RL transform requires more data than the best performing classifier to achieve the same accuracy as AR or ACF, this factor is at worst non-increasing with the series length, m, whereas the relative time taken to fit AR and ACF increases with m. We conclude that if the data is stationary and can be suitably modelled by an AR series, and if time is an important factor in reaching a discriminatory decision, then the run length distribution transform is a simple and effective transformation to use.


Related Articles

  • Forecasting the monthly inflow rate of the Palar-Porundalar dam in Tamil Nadu using SARIMA model. Chitra, K. Mano; Selvi, R. Pangayar; Mahendran, K. // Journal of Applied & Natural Science;2019, Vol. 11 Issue 2, p375 

    Dam inflow forecasting information is essential for planning and management of the dam system. Time series analysis is the most commonly employed technique to forecast the future values based on historical information. In this study, Palar-Porandalar dam in Tamil Nadu inflow series were...

  • Mischaracterising density dependence biases estimated effects of coloured covariates on population dynamics. Lindén, Andreas; Fowler, Mike; Jonzén, Niclas // Population Ecology;Jan2013, Vol. 55 Issue 1, p183 

    Environmental effects on population growth are often quantified by coupling environmental covariates with population time series, using statistical models that make particular assumptions about the shape of density dependence. We hypothesized that faulty assumptions about the shape of density...

  • BAYESIAN ESTIMATION OF CHANGE POINT IN AUTOREGRESSIVE PROCESS. Pandya, Mayuri; Bhatt, Krishnam; Thakar, HardikPandya Chetan // International Journal of Research & Reviews in Applied Sciences;Oct2012, Vol. 13 Issue 1, p41 

    Let AR(1) model be Xi = β1 X i - 1 + εi , where ei are assumed to have Normal distribution with mean 0 and variance σ12 and autocorrelation coefficient β1 but later it was found that there was a change in the system at some point of time m and it is reflected in the sequence after Xm...

  • DOES THE BIC ESTIMATE AND FORECAST BETTER THAN THE AIC? MEDEL, CARLOS A.; SALGADO, SERGIO C. // Revista de Analisis Economico;abr2013, Vol. 28 Issue 1, p47 

    We test two questions: (i) Is the Bayesian Information Criterion (BIC) more parsimonious than Akaike Information Criterion (AIC)? and (ii) Is BIC better than AIC for forecasting purposes? By using simulated data, we provide statistical inference of both hypotheses individually and then jointly...

  • Time Series Modeling for Trend Analysis and Forecasting Wheat Production of India. Dasyam, Ramesh; Pal, Soumen; Rao, Vatluri Srinivasa; Bhattacharyya, Banjul // International Journal of Agriculture, Environment & Biotechnolog;Jun2015, Vol. 8 Issue 2, p303 

    Wheat is one of the most important staple food grains of human for centuries. It has a special place in the Indian economy because of its significance in food security, trade and industry. This study made an attempt to model and forecast the production of wheat in India by using annual time...

  • A Radiated-Noise Simulator for Underwater Target Based on Vector Time Series. Yang Ju; Xinyong Wang // Applied Mechanics & Materials;2014, Issue 644-650, p4023 

    The vector time series model for simulating the underwater target radiated-noise is developed in this paper. Experimental results show that the true value lying outside the confidence interval would be a small probability event.

  • Intervention Analysis of Nigeria's Foreign Exchange Rate. MOSUGU, J. K.; ANIETING, A. E. // Journal of Applied Sciences & Environmental Management;Sep2016, Vol. 20 Issue 3, p891 

    This paper investigated the impact of Nigeria's foreign exchange rate using classical multiple regression model under the assumptions of ordinary least squares method (OLS) and intervention model using lag operator L. Monthly time series data spanning 1980:1 to 2014:12 were used and a number of...

  • APPLICATION OF SEASONAL TIME SERIES MODEL TO RAINFALL AND TEMPERATURE FORECAST. Taiwo, A. I. // Annals. Computer Science Series;2019, Vol. 17 Issue 2, p243 

    Climate change has become the foremost global challenge facing human’s existence at the moment and the effect will be multifaceted with respect to social, economy and environmental challenges. This paper built a model to forecast the significant changes in Nigerian rainfall and...

  • Modelling Road Accidents: An Approach Using Structural Time Series. Md Junus, Noor Wahida; Ismail, Mohd Tahir // AIP Conference Proceedings;2014, Vol. 1613, p228 

    In this paper, the trend of road accidents in Malaysia for the years 2001 until 2012 was modelled using a structural time series approach. The structural time series model was identified using a stepwise method, and the residuals for each model were tested. The best-fitted model was chosen based...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics