pith. sign in

arxiv: 2408.08630 · v3 · pith:YECSN2TQnew · submitted 2024-08-16 · 📊 stat.ME

Spatial Principal Component Analysis and Moran Statistics for Multivariate Functional Areal Data

Dharini Pathmanathan , Issa-Mbenard Dabo , Tzung Hsuen Khoo , Alaa Ali-Hassan , Sophie Dabo-Niang This is my paper

Pith reviewed 2026-05-23 22:22 UTC · model grok-4.3

classification 📊 stat.ME
keywords multivariate functional dataMoran's Ispatial principal component analysisareal dataspatial autocorrelationpermutation testfunctional data analysis
0
0 comments X

The pith

Multivariate functional Moran's I and mfasPCA measure spatial autocorrelation in functional areal data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops multivariate functional Moran's I to quantify spatial dependence across multiple functional variables observed on areal units. It pairs this with multivariate functional areal spatial principal component analysis (mfasPCA) to extract dominant spatial patterns from the same data. A permutation-based testing procedure is added that includes omnibus tests for positive and negative subspaces, per-component tests adjusted by Holm's method, and sequential rank-wise checks. Simulation studies and one empirical application are used to show that these tools recover known autocorrelation structures and identify meaningful components.

Core claim

Multivariate functional Moran's I, mfasPCA, and the integrated permutation testing framework accurately assess spatial autocorrelation and structural patterns in functional areal data.

What carries the argument

multivariate functional Moran's I and mfasPCA, which extend Moran's I and PCA to handle multivariate functional observations on spatial areal units while preserving the functional and spatial structure.

If this is right

  • The omnibus tests can detect dependence in both positive and negative spatial subspaces.
  • Component-wise tests with Holm adjustment control the family-wise error rate across eigencomponents.
  • The sequential rank-wise procedure orders components by strength of spatial signal.
  • The methods apply directly to empirical functional areal datasets to reveal hidden spatial structures.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same permutation framework could be reused to compare different spatial weight matrices without new theory.
  • mfasPCA scores might serve as low-dimensional inputs for subsequent spatial regression models on the same areal units.
  • Extension to non-areal domains such as point-referenced functional data would require only changes to the weight matrix definition.

Load-bearing premise

The simulation studies and empirical application are representative enough to establish the accuracy and general utility of the new statistics and testing procedures for functional areal data.

What would settle it

A new simulation study in which the proposed statistics and tests fail to detect planted spatial autocorrelation patterns of known strength in multivariate functional areal data would falsify the efficacy claim.

Figures

Figures reproduced from arXiv: 2408.08630 by Alaa Ali-Hassan, Dharini Pathmanathan, Issa-Mbenard Dabo, Sophie Dabo-Niang, Tzung Hsuen Khoo.

Figure 1
Figure 1. Figure 1: Variance explained by the top four principal components in 50 simulated data sets with n = 500 from Model 1, using fPCA and fasPCA (positive and negative scores). 16 [PITH_FULL_IMAGE:figures/full_fig_p016_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Variance explained by the top three principal components (positive) in 50 simulated data sets for rook contiguity weights: 1. red=MFPCA 2. green=mfasPCA 3. yellow=STPCA. Variance explained (%) 20 40 60 80 100 1 2 (a) ρ = 0.3 Variance explained (%) 20 40 60 80 100 1 2 (b) ρ = 0.5 Variance explained (%) 20 40 60 80 100 1 2 (c) ρ = 0.7 Variance explained (%) 20 40 60 80 100 1 2 (d) ρ = 0.9 [PITH_FULL_IMAGE:f… view at source ↗
Figure 3
Figure 3. Figure 3: Variance explained by the top three principal components (negative) in 50 simulated data sets for rook contiguity weights: 1. green=mfasPCA 2. yellow=STPCA. 17 [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Second-level administrative divisions in France based on data generated using the model in Section 3.2 for: (a) positive spatial autocorrelation and (b) negative spatial autocorrelation (rook’s contiguity weight matrix) over 101 evenly spaced time points with ρ = 0.7 for 10 variables. 0 20 40 60 80 100 0.3 0.4 0.5 0.6 x Bivariate Functional Moran (a) 0 20 40 60 80 100 0.40 0.45 0.50 0.55 0.60 x Multivariat… view at source ↗
Figure 5
Figure 5. Figure 5: Simulated (a) bivariate and (b) multivariate functional Moran’s I indices for KNN weights based on ρ = 0.7. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The male log death rates in 2010 for 28 European countries: (a) smoothed, (b) reconstructed using the matrix multiplication of the scores and principal components based on KNN weights. −2.5 −1.5 −0.5 0.5 1.5 a −0.3 −0.1 0.1 0.3 0.5 b −0.5 −0.3 −0.1 0.1 0.3 0.5 c [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The log death rates scores of the (a) first positive, (b) second positive, and (c) first negative eigenvalues of the fasPCA based on KNN weights for males from 28 European countries in 2010. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Spatial typology of Polish regions as an effect of full (a) positive and (b) negative spatial autocor￾relation (rook’s contiguity weight matrix) 20 [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Functional Moran’s I statistic curves: (a) Univariate case: log death rates for males aged 0 to 100+ in 2010 using the KNN weight matrix; (b) Bivariate case (variables 2 and 3 from [PITH_FULL_IMAGE:figures/full_fig_p021_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Polish data (variable 2 in [PITH_FULL_IMAGE:figures/full_fig_p022_10.png] view at source ↗
read the original abstract

The paper introduces a multivariate functional areal spatial principal component analysis (mfasPCA) framework, together with multivariate functional Moran's I statistics, to enable the assessment of spatial autocorrelation and dimension reduction for multivariate functional data observed over areal units. The proposed framework is spatial-functional in scope: the functional argument may represent time, age, wavelength, or another ordered continuum, while spatial dependence is introduced across areal units through a spatial weight matrix. The principal component method is defined through a Moran-type spatially weighted criterion. We propose eigenvalue-based permutation tests to assess the significance of spatially structured components. The testing framework includes omnibus tests, componentwise tests with Holm adjustment, and sequential rank-wise tests based on tail sums of eigenvalues. Simulation studies show that mfasPCA captures positive and negative spatial-functional structures and concentrates them in the leading components under the respective autocorrelation regimes. A real-data application illustrates how mfasPCA identifies spatially structured modes of multivariate functional variation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 1 minor

Summary. The paper develops multivariate functional Moran's I along with multivariate functional areal spatial principal component analysis (mfasPCA) for functional areal data. It also introduces a permutation-based testing framework that includes omnibus tests for spatial dependence in positive and negative subspaces, component-wise per-eigen tests with Holm correction for family-wise error rate control, and a sequential rank-wise testing procedure. Efficacy is asserted via comprehensive simulation studies and an empirical application demonstrating accurate assessment of spatial autocorrelation and structural patterns.

Significance. If the derivations, simulation designs, and empirical results hold, the work extends classical spatial statistics (Moran's I and PCA) to the multivariate functional areal setting and supplies a coherent testing framework; this could fill a methodological gap for data types common in environmental and geographic applications.

minor comments (1)
  1. The abstract asserts efficacy from simulations and an application but provides no quantitative details (e.g., power curves, type-I error rates, or data characteristics); the full manuscript should supply these in a dedicated simulation section with explicit design parameters.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their summary of the manuscript. The recommendation is listed as uncertain, yet the report contains no specific major comments to address. We remain available to provide clarifications or revisions should any particular concerns arise.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper introduces new methodological contributions—multivariate functional Moran's I, mfasPCA, and a permutation-based testing framework—explicitly presented as developments for functional areal data. Validation occurs via simulation studies and empirical application, which constitute independent empirical checks rather than reductions of predictions to fitted inputs or self-citations. No load-bearing steps in the abstract or described structure equate outputs to inputs by construction, self-definition, or imported uniqueness theorems; the evidentiary chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no explicit free parameters, axioms, or invented entities; the methods are presented as novel developments without detailing underlying assumptions or fitted quantities.

pith-pipeline@v0.9.0 · 5670 in / 1042 out tokens · 16095 ms · 2026-05-23T22:22:23.428760+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.