2
$\begingroup$

I'm fairly new to time series classification. I have an employee churn problem where I have quarterly data of the employees available.

Should I group my observations by time (example in table A) or by employee (example in table B)?

Table A
|---------------|---------------|---------------|---------------|
|  Employee Nr  |   Timestamp   | Other_Features|    Churned    |
|---------------|---------------|---------------|---------------|
|       1       |      1        |      --       |       0       |
|       2       |      1        |      --       |       0       |
|       1       |      2        |      --       |       1       |
|       2       |      2        |      --       |       0       |
|---------------|---------------|---------------|---------------|


Table B
|---------------|---------------|---------------|---------------|
|  Employee Nr  |   Timestamp   | Other_Features|    Churned    |
|---------------|---------------|---------------|---------------|
|       1       |      1        |      --       |       0       |
|       1       |      2        |      --       |       1       |
|       2       |      1        |      --       |       0       |
|       2       |      2        |      --       |       0       |
|---------------|---------------|---------------|---------------|
$\endgroup$
1
$\begingroup$

Your example seems to be a Panel Data rather than a time series, because there are both timestamps and "spatial" stamps (individual). However as the name states, a time series should be ordered by time continuously, that is Table B. But as this is actually a Panel Data that choice might have other consequences, for example if you want to model Dynamic Panel Data, and you have plenty time observations.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.