I am working on a data science project where I have 4 different dataframes representing 4 different metrics (let's say, met1, met2, met3 and met4). These metrics are time series and each one of them was measured on 100 different individuals.

The dataframes are structured as follow :

Indexes : Time vector

Columns : Number of individual (eg, 100 columns)

The cells contain the value of the metric at a certain time (line) for a certain individual (column).

Example for one metric :

0 1 2 ... 99
0 0.000241 0.000211 0.000125 ... 0.000227
1 0.000241 0.000248 0.000175 ... 0.000227
... ... ... ... ... ...
39 0.000211 0.000351 0.000158 ... 0.00187

Questions :

I want to build a simple regression model where the target is met1, as a function of met2 and met3 fixed (replaced by a given constant).

  1. How would one manage the data processing phase ?
  2. How can we manage features when they are dataframes and not simple vectors, to build the regression model ?

Thank you for your help.

  • $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$ Jun 25, 2023 at 11:41

1 Answer 1


It would be great to have more details about your model and objective to give a more clear answer.

As always, I would start with a good time-series exploratory data analysis. You have a good package to help you with the initial data exploration for time-series (github.com/ydataai/ydata-profiling). They include ACF and PACF plots that are useful for you to better understand the time-series lags relations, if your time-series is stationary,etc.

After that you might have a better perspective on what data preparation you need to build your model.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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