irf | Incremental Random Forest

this could help.

Plotting impulse response functions in grid format

(I can not comment yet.. So please feel free to delete this as this is not an answer.)

The problem is that deconvolve is a sort of polynomial division, it decomposes the output signal in $conv(h, x) + r$, if your signal is noisy it may give strange results. Also if the first sample in the inpulse response is small it tends to produce the exponentially growing output.

What I would do for this problem is the division of FFTs.

The essential problem here is that irf.plot is creating a figure with a single subplot on it (one row, one column). You then add a second subplot with add_subplot(211), which is telling matplotlib to arrange it in the top row of a 2-row configuration. But the original subplot is not being altered, so part of it is hidden by the new subplot (ax1).

irf.plot doesn't appear to have any options to control this at the time of creation as far as I can see.

But what we can do is modify the subplot position using GridSpec after creation (e.g. like in this answer), before adding the new subplot.

For example, something like this might work:

Finally found the solution!

I followed Dale K's hint about stripping down the code to the minimum required and came up the the conclusion that it actually was a problem about maximum query length topping higher than 8000 characters (just like AlwaysLearning suggested).

The problem happened whenever i added .Text at the very end of the query variable, causing this 8k characters limit and an error output while executing the VBA.

I solved it by calling the query variable and adding .Value at the very end of it. (str_QuerySQL.Value) and now i managed to run even queries that have 15k+ characters.

Thank you all for the hints.

Since it is a function within the vars package, you can look at the source code the same way that you did for irf.varest.

In almost all cases, you can simply drop the CTE's ORDER BY clause. Even if it were permitted syntactically (it is in other RDBMS), it has no effect on your query's result the way you wrote it.

Now, if for some reason, you absolutely have to keep it there, you can add a TOP clause, e.g. one without any effect such as TOP 100 PERCENT, i.e.:

There are a couple of issues. The first problem is that the plot method from BGVAR only exports back the 25% and 75% confidence limits. Wheras the plot also has a ribbon showing the 16% and 84% confidence limits.

To obtain these additional data points, I wrote a slightly altered version of the package's plot function, that returns both these limits in a list. There might be a simpler way to obtain these using the package functions, but I'm not familiar with BGVAR.

Your first desired result is easily achieved by resetting the aes_params after calling plot. For your second goal. There is probably an approach to manipulate the ggplot object. Instead my approach below constructs the plot from scratch. Basically I copy and pasted the data wrangling code from vars:::plot.hd and filtered the prepared dataset for the desired series:

It is always helpful to post a complete, minimal example of what you are trying to do. Without a complete example, only vague answers are possible.

You could simply do the convolutions in your model function that is wrapped by lmfit.Model, passing in the kernel array to use in the convolution. Or you could create convolution kernel and function, and do the convolution as part of the modeling process, as described for example at https://lmfit.github.io/lmfit-py/examples/documentation/model_composite.html

I would imagine that the first approach is easier if the kernel is not actually meant to be changed during the fit, but it's hard to know that for sure without more details.