From sparse to dense functional data in high dimensions: Revisiting phase transitions from a non-asymptotic perspective

Shaojun Guo,Dong Li,Xinghao Qiao,Yizhu Wang
2023-06-07
Abstract:Nonparametric estimation of the mean and covariance functions is ubiquitous in functional data analysis and local linear smoothing techniques are most frequently used. Zhang and Wang (2016) explored different types of asymptotic properties of the estimation, which reveal interesting phase transition phenomena based on the relative order of the average sampling frequency per subject $T$ to the number of subjects $n$, partitioning the data into three categories: ``sparse'', ``semi-dense'' and ``ultra-dense''. In an increasingly available high-dimensional scenario, where the number of functional variables $p$ is large in relation to $n$, we revisit this open problem from a non-asymptotic perspective by deriving comprehensive concentration inequalities for the local linear smoothers. Besides being of interest by themselves, our non-asymptotic results lead to elementwise maximum rates of $L_2$ convergence and uniform convergence serving as a fundamentally important tool for further convergence analysis when $p$ grows exponentially with $n$ and possibly $T$. With the presence of extra $\log p$ terms to account for the high-dimensional effect, we then investigate the scaled phase transitions and the corresponding elementwise maximum rates from sparse to semi-dense to ultra-dense functional data in high dimensions. Finally, numerical studies are carried out to confirm our established theoretical properties.
Statistics Theory
What problem does this paper attempt to address?
### What problem does this paper attempt to solve? This paper primarily explores the issue of functional data analysis from sparse to dense in high-dimensional functional data. Specifically: 1. **Non-parametric Estimation**: - Conducts non-parametric estimation of the mean function and covariance function in functional data. - Uses local linear smoothing methods to handle discretely sampled and noisy data. 2. **Phase Transition Phenomenon**: - Studies the phase transition phenomenon and its convergence rate under different sampling frequencies (sparse, semi-dense, ultra-dense). - Proposes phase transition analysis based on the ratio of average sampling frequency to sample size and extends it to high-dimensional cases. 3. **Non-asymptotic Properties**: - Derives concentration inequalities for local linear smoothers to obtain non-asymptotic error bounds. - These results reveal L2 convergence rates and uniform convergence rates similar to those of Zhang and Wang (2016), and demonstrate the impact of introducing additional log p terms in high-dimensional cases. 4. **Convergence Properties in High Dimensions**: - Investigates the element-wise maximum convergence rate of the mean function and covariance function estimation in high-dimensional settings. - Provides an extended theoretical framework for functional data from sparse to semi-dense to ultra-dense, and validates its effectiveness through numerical simulations. In summary, this paper aims to provide a systematic non-asymptotic theoretical analysis for non-parametric estimation of high-dimensional functional data, and to reveal the phase transition phenomenon and optimal bandwidth selection under high-dimensional effects.