1
$\begingroup$

I am wondering what would be the correct ML approach in order to predict the upcoming value of a time serie based on the previous behaviours of various time series for the same period.

I have a dataset in the form of:

TS name Day1 Day2 ... Day50 Target-Day51
TS 1 5 13 ... 16 12
TS 2 8 18 ... 9 16
... 12 2 ... 13 4
TS 4000 3 7 ... 4 10

Imagine that a new row will be in the following form and I want to predict the target day:

TS name Day1 Day2 ... Day50 Target-Day51
TS 4001 3 22 ... 48 XX

Is this a time-series approach? A regression one? A Multivariate Time series one? Can you suggest some algorithms which could work here?

Improve this question
$\endgroup$
1
  • $\begingroup$ Looks like a Vector Autoregression (VaR) model may be appropriate. $\endgroup$
    – user2974951
    Nov 26, 2021 at 12:16

1 Answer 1

Reset to default
0
$\begingroup$

There is a nice overview of possible techniques/models for R-Packages:

https://cran.r-project.org/web/views/TimeSeries.html

Especially have a look at the section "Multivariate Time Series Models" (e.g. Vector autoregressive (VAR) models). However, other approaches could be interesting for you as well.

In case you have a sufficient amount of data, you could also look at neural nets, i.e. with LSTM layers. Find a instructive example here:

https://keras.io/examples/timeseries/timeseries_weather_forecasting/

Improve this answer
$\endgroup$
4
  • $\begingroup$ Dear @peter, thank you so much for your answer. Helped a lot. I have a follow-up question here in case you can help me more. My final target is not to predict the target day of the above time serie (TS4001) based on how it is affected from all the others (like in weather prediction). My target is to predict the TS4001 by matching its pattern with the 'same pattern' time series from the train test. $\endgroup$
    – DS_tech_member
    Dec 20, 2021 at 9:35
  • $\begingroup$ E.g. 4001 is coming and has seasonality. I want the algorithm to match this pattern with the similar patterns time series from the train dataset and to predict the target day based on the way that the 51 day turns out for the train set. Please let me know if something is not clear enough. $\endgroup$
    – DS_tech_member
    Dec 20, 2021 at 9:35
  • $\begingroup$ I guess this is a separate question and I do not really understand the problem. Maybe open up a new question on this. $\endgroup$
    – Peter
    Dec 20, 2021 at 17:40
  • $\begingroup$ Thanks @peter . I created a new question here: datascience.stackexchange.com/questions/106285/… $\endgroup$
    – DS_tech_member
    Dec 21, 2021 at 10:06

Your Answer

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

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