×

Group Lasso for structural break time series. (English) Zbl 1367.62251

Summary: Consider a structural break autoregressive (SBAR) process \[ Y_t=\mathop{\sum}\limits_{j=1}^{m+1}[\beta^T_jY_{t-1}+\sigma(Y_{t-1},\ldots,Y_{t-q})\varepsilon_t]I(t_{j-1}\leq t <t_j), \] where \(Y_{t-1}=(1,Y_{t-1},\ldots,Y_{t-p})^T\), \(\beta_j=(\beta_{j0},\ldots,\beta_{jp})^T\in \mathbb{R}^{p+1}\), \(j = 1,\ldots, m + 1\), \(\{t_1,\ldots, t_m\}\) are change-points, \(1 = t_0 < t_1 <\cdots < t_{m+1} = n + 1\), \(\sigma(\cdot)\) is a measurable function on \(\mathbb{R}^q\), and \(\{\varepsilon_t\}\) are white noise with unit variance. In practice, the number of change-points \(m\) is usually assumed to be known and small, because a large \(m\) would involve a huge amount of computational burden for parameters estimation. By reformulating the problem in a variable selection context, the group least absolute shrinkage and selection operator (LASSO) is proposed to estimate an SBAR model when \(m\) is unknown. It is shown that both \(m\) and the locations of the change-points \(\{t_1,\ldots,t_m\}\) can be consistently estimated from the data, and the computation can be efficiently performed. An improved practical version that incorporates group LASSO and the stepwise regression variable selection technique are discussed. Simulation studies are conducted to assess the finite sample performance.

MSC:

62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH)
62M07 Non-Markovian processes: hypothesis testing
62J07 Ridge regression; shrinkage estimators (Lasso)
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] DOI: 10.1080/01621459.1998.10473808 · doi:10.1080/01621459.1998.10473808
[2] Adak S., Time-Frequency Representations: Algorithms and Applications pp 261– (1996)
[3] Andreou E., Handbook of Financial Time Series pp 839– (2008)
[4] DOI: 10.2307/2951764 · Zbl 0795.62012 · doi:10.2307/2951764
[5] DOI: 10.1162/003465397557132 · doi:10.1162/003465397557132
[6] DOI: 10.2307/2998540 · Zbl 1056.62523 · doi:10.2307/2998540
[7] DOI: 10.1002/jae.659 · doi:10.1002/jae.659
[8] DOI: 10.1016/0020-0255(83)90009-9 · Zbl 0568.94004 · doi:10.1016/0020-0255(83)90009-9
[9] DOI: 10.1214/08-AOS620 · Zbl 1173.62022 · doi:10.1214/08-AOS620
[10] DOI: 10.1214/07-AOS558 · Zbl 1155.62034 · doi:10.1214/07-AOS558
[11] DOI: 10.1002/9781118032466 · Zbl 1284.62008 · doi:10.1002/9781118032466
[12] DOI: 10.1198/016214505000000745 · Zbl 1118.62359 · doi:10.1198/016214505000000745
[13] DOI: 10.1198/jasa.2010.tm09181 · Zbl 1388.62211 · doi:10.1198/jasa.2010.tm09181
[14] Kim C.J., State-Space Models with Regime Switching (1999)
[15] DOI: 10.1016/j.jvolgeores.2007.10.006 · doi:10.1016/j.jvolgeores.2007.10.006
[16] Mendenhall W., A Second Course in Statistics: Regression Analysis (7th ed.) (2011)
[17] DOI: 10.1198/016214501753168244 · Zbl 1018.62080 · doi:10.1198/016214501753168244
[18] DOI: 10.1016/0020-0255(83)90007-5 · Zbl 0584.62154 · doi:10.1016/0020-0255(83)90007-5
[19] DOI: 10.1198/jasa.2010.tm10103 · Zbl 1390.62184 · doi:10.1198/jasa.2010.tm10103
[20] Tibshirani R., Journal of the Royal Statistical Society, Series B 58 pp 267– (1996)
[21] DOI: 10.1111/j.1467-9868.2005.00490.x · Zbl 1060.62049 · doi:10.1111/j.1467-9868.2005.00490.x
[22] DOI: 10.1002/0471704091 · Zbl 1068.62077 · doi:10.1002/0471704091
[23] DOI: 10.1111/j.1467-9868.2005.00532.x · Zbl 1141.62030 · doi:10.1111/j.1467-9868.2005.00532.x
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.