oreum-industries / oreum_core Goto Github PK

Github? https://github.com/features/packages

Gemfury: service: https://gemfury.com/l/pypi-server

DIY: https://testdriven.io/blog/private-pypi/

nbconvert 5.6.1

Regular latex template has images too large:

\adjustimage{max size={0.9\linewidth}{0.9\paperheight}}{((( filename )))}

I want to override using classic_mod.tplx, something like

((*- block figure -*))
    ((( super() )))
    %\renewcommand{\adjustimage}{max size={0.5\linewidth}{0.5\paperheight}}{((( filename )))}
((*- endblock figure -*))

but nothing I try works. Instead have manually hacked

/Users/jon/opt/anaconda3/envs/freberg_trk/lib/python3.8/site-packages/nbconvert/templates/latex/document_contents.tplx

with line:

\adjustimage{max size={0.9\linewidth}{0.3\paperheight}}{((( filename )))}

which is poor form!

Improve summary e.g. https://github.com/pymc-devs/pymc/releases/tag/v5.0.1
and include comparison e.g. v0.4.7...v0.4.8

useful for e.g. freq defined as claims / $exposure which is zero-inflated

consider zero-inflated poison / neg-binomial, but these are discrete and freq defined over $exposure is continuous, so prefer zero inflated something else... perhaps Gamma:

e.g.

• https://stackoverflow.com/questions/42409761/pymc3-nuts-has-difficulty-sampling-from-a-hierarchical-zero-inflated-gamma-mode
• http://xcelab.net/rmpubs/Mcelreath%20Koster%202014.pdf

see also:

• https://discourse.pymc.io/t/zero-inflated-poisson-log-lik/2664
• https://stats.stackexchange.com/questions/451091/pymc3-applying-a-zero-inflated-poisson-model

Create calc_2_sample_delta_prop

Calculate 2-side sample delta difference between arrays row-wise so that we can make a statement about the difference between a test array a reference array how different arr is from arr_ref

    Basic algo
    ---------- 

    for each row i in arr:
        for each row j in arr_ref:
            do: d = arr[i] - arr_ref[j]
            do: q = quantiles[0.03, 0.97](d)
            do: b_i = q > 0
            if: sum(b_i) == 2:
                we state "94% of arr[i] > arr_ref[j], substantially larger"
            elif: sum(b_i) == 1:
                we state "not different"                
            else (sum(b_i) == 0):                
                we state "94% of arr[i] < arr_ref[j], substantially smaller"                    
        do: prop = unique_count(b) / len(b)

Then we can state "prop arr[i] larger, same, smaller than arr_ref"

https://docs.github.com/en/packages/quickstart

similar to #1

simple one and fixed in new branch