The well-established practice of time series analysis involves estimating deterministic, non-stationary trend and seasonality components followed by learning the residual stochastic, stationary components. Recently, it has been shown that one can learn the deterministic non-stationary components accurately using multivariate Singular Spectrum Analysis (mSSA) in the absence of a correlated stationary component; meanwhile, in the absence of deterministic non-stationary components, the Autoregressive (AR) stationary component can also be learnt readily, e.g.

Abstract-- Time-inconsistent preferences, where agents favor smaller-sooner over larger-later rewards, are a key feature of human and animal decision-making. Quasi-Hyperbolic (QH) discounting provides a simple yet powerful model for this behavior, but its integration into the reinforcement learning (RL) framework has been limited. We make two primary contributions: (i) we formally characterize the structure of the optimal policy, proving for the first time that it reduces to a simple one-step non-stationary form; and (ii) we design the first practical, model-free algorithms for both policy evaluation and Q-learning in this setting, both with provable convergence guarantees. Our results provide foundational insights for incorporating QH preferences in RL. Reinforcement learning (RL) [12] provides a powerful framework for sequential decision-making, where an agent interacts with an environment to maximize long-term cumulative rewards.

The well-established practice of time series analysis involves estimating deterministic, non-stationary trend and seasonality components followed by learning the residual stochastic, stationary components. Recently, it has been shown that one can learn the deterministic non-stationary components accurately using multivariate Singular Spectrum Analysis (mSSA) in the absence of a correlated stationary component; meanwhile, in the absence of deterministic non-stationary components, the Autoregressive (AR) stationary component can also be learnt readily, e.g. However, a theoretical underpinning of multi-stage learning algorithms involving both deterministic and stationary components has been absent in the literature despite its pervasiveness. We resolve this open question by establishing desirable theoretical guarantees for a natural two-stage algorithm, where mSSA is first applied to estimate the non-stationary components despite the presence of a correlated stationary AR component, which is subsequently learned from the residual time series. We provide a finite-sample forecasting consistency bound for the proposed algorithm, SAMoSSA, which is data-driven and thus requires minimal parameter tuning.