Evaluating time-series in spatial dimension is about including spatial-autocorrelation. We evaluate how autocorrelation in noise (residuals), confounders, and signal affect estimates at two additional levels of autocorrelation and two levels of data availability
Process:
generate spatial random field with specified level of autocorrelation
randomly place pixels / [ place pixels in a grid ]?
induce autocorrelation by multiplying
1-2 moderate-large N, 50/50 mismatch, moderate noise
Idea:
Evaluating time-series in spatial dimension is about including spatial-autocorrelation. We evaluate how autocorrelation in noise (residuals), confounders, and signal affect estimates at two additional levels of autocorrelation and two levels of data availability
Process:
1-2 moderate-large N, 50/50 mismatch, moderate noise