# Modelling multi vector time series in PyMC3

**URL:** <https://discourse.pymc.io/t/modelling-multi-vector-time-series-in-pymc3/6215>\
**Category:** Questions\
**Created:** [November 5, 2020, 9:13am UTC](https://discourse.pymc.io/t/modelling-multi-vector-time-series-in-pymc3/6215 "2020-11-05T09:13:51Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![IvanSanych](https://avatars.discourse-cdn.com/v4/letter/i/35a633/32.png) [@IvanSanych](https://discourse.pymc.io/u/IvanSanych)\
**Post date:** [November 5, 2020, 9:13am UTC](https://discourse.pymc.io/t/modelling-multi-vector-time-series-in-pymc3/6215/1 "2020-11-05T09:13:51Z")

</div>

I apologize for being vague, I’ll try to explain myself as well as I can.

I try to figure out whether PyMC3 can help me in this problem

We have ~20 vectors of ~100 measurements each. These vectors represent ~20 biological systems. We let each system evolve over time, and take the ~100 measurements at fixed intervals. In the end, each of the ~20 systems has 15 equally spaced measurements.

We can think of the evolution of the system as a 100x100 matrix that defines the value transition from one vector element to another.

Our goal is to find a model that will shed the light on this 100x100 matrix and, possibly, to predict the final state of the vector, from its initial values.

---

<div class="post-metadata">

**Author:** ![junpenglao](https://yyz2.discourse-cdn.com/flex036/user_avatar/discourse.pymc.io/junpenglao/32/8_2.png) [@junpenglao](https://discourse.pymc.io/u/junpenglao)\
**Post date:** [November 5, 2020, 9:29am UTC](https://discourse.pymc.io/t/modelling-multi-vector-time-series-in-pymc3/6215/2 "2020-11-05T09:29:55Z")

</div>

I guess it is some kind of state space model? I think it is possible to write the timeseries realization using a theano.scan