Reputation: 11
Let's assume we have some time-series data where every person (sample) has measurements for 2 variables (Var1, Var2) for three timepoints. Following is a dummy representation of the data.
Is there a mathematical/statistical approach to quantify/detect if a variable (e.g., Var1) varies primarily across timestamps or across persons/samples?
I am initially thinking to take average of a variable across samples for each time point and then take the variance of these average values. Similarly, taking average of a variable across tiemstamps for each sample, and then taking variance of the average values. Then checking which variance has a higher magnitude. But this is just an intuitive speculation.
I wonder if there is any mathematical or statistical technique to detect/quantify the source(timestamp/sample) of variation for each variable.
Thanks in advance!
Upvotes: 1
Views: 30