@mmcdara 

Besides the execution time, something else will worry me. When I import more amount of data as a DataFrame, I see a distinct delay in characters refresh. It looks like the typed characters appear with an apparent and frustrating delay of even a few seconds.

Also, you're right, DataFrame logic is quite similar to native R.
So I really regret that it doesn't work as well in Maple as it does in R.
In certain types of analysis, DataFrame would be very helpful.

wzelik

in relation to my question...

Data from Excel: 150 000 rows x 13 columns (in attachment).

Example task: count the "single" value in the fifth column (name: "Married_Single").

Import via ImportMatrix() - 0.813 s

Import to DataFrame via Import() - 41.563s 

wzelik

150kX13_LOAN.xlsx

Maple Flow Team,

Maple Flow 2022.2 - big step forward in performance.

I keep my fingers crossed for the development of Maple Flow.

wzelik

@mmcdara 
Including you😀

I appreciate

wzelik

@acer 

It's a little bit complicated but really fast.

Calculation on my XPS 9710 in Maple 2022.2  takes 0.2 s for 50 000 samples.

Thank you.

wzelik

@mmcdara 

It seems that your procedure proposition is much more efficient than the preliminary mine.

The execution time is almost twice as short as mine (11.6 s for 50 000 samples), but still, the calculation time is quite far from the solution in Mathcad Prime 8.0.

Anyway, thank you for your effort

@mmcdara 

Thank you,

As you are probably aware, it was not my intention to use the well-known and documented conjugate prior method.

In fact, I wanted to reproduce by Maple the results I got a few months ago using the mcmcPack packages in R (https://cran.r-project.org/web/packages/MCMCpack/), where I used the MCbinomialbeta function to compute the posterior distribution.

Of course, I know the exact solution you gave, Beta(23,3), from the method of the conjugate prior. So I could easily compare the results. At the same time, I watched the trace plot and tried to observe the ratio calculated by following the procedure:

wzelik

@Carl Love 

Thank you for your prompt reply.

I have tested, samp:=mcmcBinomialBeta(21, 1, 4, 2, 0.5, 50000, 500, 0.1) # timeCalc = 20.56 s

a, b - parameters of Beta distribution, type: Integer,

n, k - parameters of Binomial distribution, type: Integer,

init - initial value, type float,

N - number of samples, type Integer,

burn - reject first burn value samples, type: Integer,

thin - adjust results if needed.

with(Statistics):

Mean(samp); # 0.884957164100278
Quantile(samp, 0.025); # 0.742018607354109
Quantile(samp, 0.975);0.975124932511784

plots:-pointplot(<seq([n, samp(n)], n = 1 .. numelems(samp))>, axes = boxed, font = [Helvetica, Bold, 10], gridlines = true, colour = red, size = [500, 300], symbol = diagonalcross);

Histogram(samp, averageshifted = 4, style = polygon, color = "LightSlateGrey", font = [Helvetica, Bold, 10])

Thank you once again.

wzelik

@Preben Alsholm 

Thank you. It seems that your nested solution works very well.

wz

@Carl Love 

wz

Hi,

Question is the same...

...but I saw that previous version of test.mw was not correct.

I need value from internal procedure of module.

wz

test.mw

@Carl Love 

please edit startup code

wz

@Preben Alsholm

Thank you very much.

My target is, to solve multilayer (different thermal diffusivity for each layer) heat transfer problem by reducing PDE to ODE form, ans solve system of ODE describing by matrix way as you wrote

As you see I have had more basic problem, than I expect :-)

Is there a way to avoid this instability and to solve this problem for each t?

wzel

Maybe problem is conected with yestardays update of MaplePrimes?

This time I put the contents...

Download HT_one_layer.mw

HT_one_layer.mw

Hi,

I'm sorry but I don't know why "404" error appears.

 Trying once again,

wzel