Forecasting Compositional Time Series: A State Space Approach

A framework for the forecasting of composite time series, such as market shares, is proposed. Based on Gaussian multi-series innovations state space models, it relies on the log-ratio function to transform the observed shares (proportions) onto the real line. The models possess an unrestricted covariance matrix, but also have certain structural elements that are common to all series, which is proved to be both necessary and sufficient to ensure that the predictions of shares are invariant to the choice of base series. The framework includes a computationally efficient maximum likelihood approach to estimation, relying on exponential smoothing methods, which can be adapted to handle series that start late or finish early (new or withdrawn products). Simulated joint prediction distributions provide approximations to the required prediction distributions of individual shares and the associated quantities of interest. The approach is illustrated on US automobile market share data for the period 1961–2013.

[1]  Kesten C. Green,et al.  Competitor-Oriented Objectives: Myth of Market Share , 2006 .

[2]  Rajkumar Venkatesan,et al.  Forecasting category sales and market share for wireless telephone subscribers: a combined approach , 2002 .

[3]  T. Smith,et al.  The time series analysis of compositional data , 1998 .

[4]  Jane M. Fry,et al.  Compositional data analysis and zeros in micro data , 2000 .

[5]  Ashton de Silva,et al.  A multivariate innovations state space Beveridge Nelson decomposition , 2009 .

[6]  Peter J. Danaher,et al.  Econometric models for forecasting market share , 2001 .

[7]  Rob J Hyndman,et al.  Stochastic population forecasts using functional data models for mortality, fertility and migration , 2008 .

[8]  John Aitchison,et al.  The Statistical Analysis of Compositional Data , 1986 .

[9]  Josep Antoni Martín-Fernández,et al.  Compositional VARIMA Time Series , 2011 .

[10]  J. Ord,et al.  Estimation and Prediction for a Class of Dynamic Nonlinear Statistical Models , 1997 .

[11]  Rob J. Hyndman,et al.  Forecasting with Exponential Smoothing , 2008 .

[12]  A. Zellner.  An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias , 1962 .

[13]  L Nyström,et al.  Statistical Analysis , 2008, Encyclopedia of Social Network Analysis and Mining.

[14]  A. Raftery,et al.  Time Series of Continuous Proportions , 1993 .

[15]  Rob J. Hyndman,et al.  Optimal combination forecasts for hierarchical time series , 2011, Comput. Stat. Data Anal..

[16]  H. Akaike.  A new look at the statistical model identification , 1974 .

[17]  J. Aitchison,et al.  Compositional Data Analysis: Where Are We and Where Should We Be Heading? , 2003 .

[18]  Terence C. Mills,et al.  Forecasting compositional time series , 2010 .

[19]  Rob J Hyndman,et al.  Another look at measures of forecast accuracy , 2006 .

[20]  Rob J Hyndman,et al.  The vector innovations structural time series framework , 2010 .

[21]  Peter Schmidt,et al.  Estimation of seemingly unrelated regressions with unequal numbers of observations , 1977 .