General Discussions

Discuss a wide range of topics! Questions about the Alteryx Platform should be directed to the appropriate Product discussion forum.

Problems and Errors with time series using covariates

Highlighted
5 - Atom

I tell you, I'm new using Alteryx, I´m using it because I'm a Computer Engineering student and with Alteryx I do my thesis to graduate.

Now I am going to talk about the problem that I have, it turns out that I am making predictions with covariates (wich are 19), to make the prediction of the enrollment that is university careers, I made a macro in witch I work arima and with prognosis, but when I use the covariates, some work and others do not, this is when I select them all (I know they do not work because they throw me an error), but if I select only one covariate and I am testing one by one, each one only if they work, what does not I undestarnd why when I select them all at once they do not work.

The covariates are all of the same type, since with macro I get errors, I try to solve the problem using the new tool, the Ts Model Factory with the Ts Forecast Factory, but the same thing happens again when I try to make the forecast.

 

The following images show the errors when wanting the same result in different processes.

 

These are the errors with the macroThese are the errors with the macroThese are the errors using Ts Model Factory with Ts Forecast FactoryThese are the errors using Ts Model Factory with Ts Forecast Factory

(Try to try two forms, the onces I mentioned but both gave me an error and I need to use the 19 covariates)

 

To consider:

 -Target variable:

      -total_matri

-Covariates:

      -t_mujeres_c, t_hombres_c, m_mujeres_p, m_hombres_p, m_mujeres_c, m_hombres_c, vacantes_s1, m_tot_primero, tipoe_cienhum_ tipoe_tecpro, tes_municipal, tes_part_sub, tes_part_pag, tes_corp_adm, total_titulados, dur_estudio_c, dur_total_c, valor_matri, valor_aran

 

I will attach the projects to see if they can help me, please, because I have tried to solve it in many ways and I have not succeeded. When I execute both forms of solution, they work, but it is because I do not have all the covariates selected in Arima, I try to select the others and will see what I am saying, I need an urgent solution. (The data is an example)

 

 

Highlighted
Alteryx Certified Partner

Not totally sure what's going on, but I know that you can't use the covariates with the Model Factory.  I don't think the Model Factory would serve well in this case.  The model factor is works great when you need a series for a variety of things.  For instances, I needed a Time Series for each of our 300+ medical clinics.  I needed to know how many medical visits each center would get for the next year.  TS Model Factory was able to produce a seperate model for each of my 300+ medical clinics.

Highlighted
5 - Atom

hi @

but I also probe it using macro with arima with covariants and  forecasts covariant, and in all the ways I get the same errors.
Regarding your model factory, I was testing with only 2 university degrees to be able to find the solution to the fastest error, but the file contains 2778 university degrees that should go to the time series, if you need them to look at it