Visualize autocorrelation in multiple timeseries one dataset

Asked Jun 15 '20 at 11:39

Active Jun 15 '20 at 12:36

Viewed 206 times

Background

Performing multiple linear regression analysis on dataset containing 21 time-series of 4 data points. Each data points represents 9 variables.

R² adjusted is not decreasing rapidly after adding more variables.

Problem statement

Since the 4 data points are a static window of a longer time-series I am thinking autocorrelation between the data-points is the problem.

Question

How to visualize or estimate if there is autocorrelation inside the different time-series?

What I have tried:

I think ACF of the residuals is not working since I have 21 time-series.

edited Jun 15 '20 at 12:36

Thalassophile

asked Jun 15 '20 at 11:39

bart12341234

Linear regression isn't what you want when you have a time series. You seem to have lots of short time series, perhaps you want to look at longitudinal data analysis? https://en.m.wikipedia.org/wiki/Longitudinal_study – jcken Jun 15 '20 at 11:47
Regarding whether you should pay attention to $R^2_{adj.}$ (perhaps only slightly relevant): "Justification for and optimality of $R^2_{adj.}$ as a model selection criterion". The short answer is, there is no answer justifying that in the thread. – Richard Hardy Jun 15 '20 at 13:11

0 Answers0