Is it possible to use time-based data?

General Discussion
,
#1

Hello and thank you everyone for this beautiful forum.

I’m learning tensorFlow but now i’ve a doubt.

Is it possible to use tensorflow in order to use time-dependant data?

I give you an example: i want to make a very simple model (i’m a beginner) about a resistance where current is flowing. Output should be resistance temperature.

Baseically input data should be like:

  • [timestamp, ampere] Output should be [temperature] (we’re assuming that ambient temperature doesn’t matter)

so every data should be like:
[1 sec, 10A] [20°C]
[2 sec, 5 A] [21°C]

and so on. I have thousand of these datas. Basically output temperature depends of what happened before due to temperature inertia.

What kind of Neural network do you think i should use? I think RNN but i’n but i’m not sure about it.

Thank you very much and best regards

#3

Hi @Paul

Welcome to the TensorFlow Forum!

You can refer few of the TF tutorials for such scenario and for time series data like Time series forecasting and Working with RNNs which might be helpful to you in this issue. Thank you.