How to interpret ACF plot y-axis scale in R
r, statistics
Solution
The scale is from -1 to 1 because it is the correlation coefficient. From the graph we can see the lags do not have significant effect (within the bounds - cannot tell them from being zero). The ACF function says if the current value depends consistently on previous values (the lags). So you see that the only value is the spike at `lag 0`. ( Try to think what it means) so your hourly values are independent of each other if only trying to explain them with themselves (the autocorrelation property).
Problem
I have created a zoo time series object for a subset of data that I have. The data is evenly spaced in hourly intervals but it is a weakly regular time series according to the R-zoo documentation (ie. there are some time points that are missing in the data) ``` df <- read.table("Desktop/AutoCorrelation_hourly_testdata.csv", header = TRUE, sep = ",") sub <- df[df$sid == 59, ] sub <- sub[!duplicated(sub[c("binstart_date", "binstart_time")]), ] library(zoo) z <- read.zoo(sub, sep = ",", header = TRUE, index = 1:2, tz = "", format = "%Y-%m-%d %H:%M:%S") acf_C_duration = acf(coredata(z$C_duration), na.action = na.pass, plot = TRUE) ``` I get the following plot: My question is what exactly is the scale of the lag on the y-axis if my data is in hourly intervals? What would be the best way for me to interpret this plot? I'm not that experienced with statistics or R, but I am trying to determine if this data set seems to follow a rhythm or have any underlying pattern/periodicity. From researching online it seems as though the R autocorrelation function could do this for me. If you think I should be using any other method please let me know. Thanks!