How to create a script in Matlab

I have the following codes: I want to bin the price data in 4 bins per month and to calculate the mean and std of demand that belong in each bin by month and year.

Can you help me to create a function that contain (the bin, month, year) as a script?? I am repeating the same code for each month, I need a structure that calculate

them very fast and save all the results in one table.

Price = Data1(:,1 );
Demand = Data1(:,2 );
tv = (datetime([2006,1,1]):datetime([2018,12,31])).';
[Ye,M,D] = ymd(tv);
DT = timetable(Price,Demand, Ye, M, D, 'RowTimes', tv );
MarchDataPD = DT(DT.M==3,{'Price','Demand'});
  topEdge1=max(P.Price(1:31));
  botEdge1=min(P.Price(1:31));
  NumBins=4;
   binEdges1=linspace(botEdge, topEdge, NumBins+1);
   [h, WhichBin]= histc(P.Price(1:31),binEdges1);
   for i = 1:NumBins
       flagBinMembers=(WhichBin==i);
       binMembers=d.Demand(flagBinMembers);
       binMean1(i)=mean(binMembers);
       binstd1(i)=std(binMembers);
   end
topEdge2=max(P.Price(32:62));
  botEdge2=min(P.Price(32:62));
  NumBins=4;
   binEdges2=linspace(botEdge, topEdge, NumBins+1);
   [h, WhichBin]= histc(P.Price(32:62),binEdges2);
   for s = 1:NumBins
       flagBinMembers=(WhichBin==s);
       binMembers=d.Demand(flagBinMembers);
       binMean2(s)=mean(binMembers);
       binstd2(s)=std(binMembers);
   end

Answer 1

[Guillaume]

It's a bit puzzling that you are aware of timetables and yet resort to deprecated functions such as histc (which has no concept of dates) for the rest of your code, instead of using timetables functions which would make your task trivially easy.

Even the way you construct your timetable is more complicated than it should:

%Data1: a 2 column matrix of price and demand:
DT = array2timetable(Data1, 'VariableNames', {'Price', 'Demand'}, 'RowTimes', datetime(2006, 1, 1):dateime(2018, 12, 31));

There is no point in storing Ye, M, D separately. It's already stored as part of the time.

Once you have a timetable, you can easily resample it at regular or irregular intervals using retime. However retime doesn't do

%DT: a timetable
meanDT = retime(DT, 'regular', 'mean', 'TimeStep', caldays(14));  %discussion about timestep coming up
stdDT = retime(DT, 'regular', @std, 'TimeStep', caldays(14));
meanDT.Properties.VariableNames = {'MeanPrice', 'MeanDemand'};
stdDT.Properties.VariableNames = {'StdPrice', 'StdDemand'};
meanstdDT = [meanDT, stdDT]

Instead of using two retime, you could use groupsummary instead:

meanstdDT = groupsummary(DT, 'Time', caldays(14), {'mean', 'std'})

Now you say you want 4 bins per month. What does that mean exactly. 4 equal bins per month means that the edges fall at odd times during the days instead of falling at the end of days. That doesn't sound like something you'd want. In the above, I've used bins of 14 days but both retime and groupsummary support a bin vector instead. To construct a bin vector with 4 equal bins per month, the simplest is probably:

monthvector = DT.Time(1) : calmonths(1) : DT.Time(end) + calmonths(1);
binvector = fillmissing(reshape([monthvector; NaT(3, numel(monthvector))], [], 1), 'linear')

and then

meanstd = groupsummary(DT, 'Time', binvector, {'mean', 'std'})

[gjashta]

Thank you for your help!

I want 4 bins per month, where each bin is the average of a specific price range,but not bins of 14 days. So the number of observations to not be the same in each bin.

For example:

In Data1, the minimum price is -0.5$ and the maximum is 654$. So, by choosing four different ranges of prices for example -0.5$-50$, 50$-84$,84$-100$, and 100$-654$. Then each bin to be the average of those (the sum of all prices in that range / the number of the prices that belong in the same range). I don't know, if I was clear enough?

[Guillaume]

Oh! I completely misunderstood the binning then. However, it's still the same method:

%Data1: a 2 column matrix of price and demand:

%convert Data1 to timetable
DT = array2timetable(Data1, 'VariableNames', {'Price', 'Demand'}, 'RowTimes', datetime(2006, 1, 1):datetime(2018, 12, 31));

%calculate the edges of the 4 price bins
[~, pricedges] = discretize(DT.Price, 4);

%calculate mean and standard deviation of Price and Demand per month and per price bin:
monthlybinnedprice = groupsummary(DT, {'Time', 'Price'}, {'month', pricedges}, {'mean', 'std'})

[gjashta]

Sorry for the delayed answer!

I run the code but pricedges has only few observation of demand most of the time.

[Guillaume]

So, I need a better binning way to have reasonable observations that belong to 4 bins of the month

Well, you're free to use whichever binning method you want. As I've no idea why you're doing this, I've nothing to suggest. Perhaps you shoud bin according to demand instead of price but the binning method is for you to decide.

In the end, whichever binning method you'd still pass the bin edges to groupsummary.

[gjashta]

Thank you Guillaume! And sorry because as a beginner in using Matlab, I don't express the question correctly.