Improve part of code using a loop
情報
This question is locked. 編集または回答するには再度開いてください。
古いコメントを表示
I have an annual table X, where I want to calculate the five maximum values for each month. My code works but I need to also apply it for other variables so I am wondering how to shrink it a bit using a loop. So far i splitted the initial annual table (X) into months like this:
X_January = table(Dates{1,1},X{1,1});
X_February = table(Dates{2,1},X{2,1});
X_March = table(Dates{3,1},X{3,1});
X_April = table(Dates{4,1},X{4,1});
X_May = table(Dates{5,1},X{5,1});
X_June = table(Dates{6,1},X{6,1});
X_July = table(Dates{7,1},X{7,1});
X_August = table(Dates{8,1},X{8,1});
X_September = table(Dates{9,1},X{9,1});
X_October = table(Dates{10,1},X{10,1});
X_November = table(Dates{11,1},X{11,1});
X_December = table(Dates{12,1},X{12,1});
%Max 5 values for January
[Col,idx] = maxk(X_January{:,1},5);
T= X_January(idx,:)
Any ideas how to make a loop?
2 件のコメント
Dyuman Joshi
2023 年 2 月 4 日
編集済み: Dyuman Joshi
2023 年 2 月 4 日
How do you want to store your final output?
All values in a numeric array (12x5 or 5x12 )? If otherwise, please specify.
Rena Berman
2025 年 1 月 13 日
(Answers Dev) Restored edit
採用された回答
One approach —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Values = randn(numel(date_time), 5);
T1 = [table(date_time) array2table(Values)]
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
end
Month{1} % January
Month{12} % December
Please do not use numbered variables! Simply store the results in an array.
.
7 件のコメント
As always, my pleasure!
It is being stored in a variable (cell array) that I call ‘Month’ that has 12 elements, one for each month. This is much easier to work with than numbered variables. Just address them as I do here to get the values for each month —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
end
Month{1} % January
Month{12} % December
The addressing will be slightly different for a timetable, since there would be only one variable in a timetable version of ‘T1’.
[~,idx] = maxk(M{:,2},1); % Index Returns 5 Highest Values Of First Variable
The rest of the code is unchanged.
.
Star Strider
2023 年 2 月 5 日
As always, my pleasure!
Yes.
Try something like this —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
VN = Month{k}.Properties.VariableNames;
[Y,M,D] = ymd(Month{k}.date_time);
Month{k} = addvars(Month{k},D,M,Y,'AFter','date_time');
Month{k}.Properties.VariableNames = {VN{1},'D','M','Y',VN{2}}; % Necessary, Since They Show Up As 'Var2' etc. Otherwise
end
Month{1} % January
Month{12} % December
I chose to keep the ‘date_time’ variable here (and I do not recommend removing it), since it could have time information that may become necessary in other applications. If you want to remove the 'date_time' variable, use the removevars function.
.
As always, my pleasure!
Yes!
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
VN = T1.Properties.VariableNames;
[Y,M,D] = ymd(T1.date_time);
DN = day(T1.date_time,'dayofyear');
T1 = addvars(T1,D,M,Y,DN,'AFter','date_time');
T1.Properties.VariableNames = {VN{1},'D','M','Y','Day Number',VN{2}}; % Necessary, Since They Show Up As 'Var2' etc. Otherwise
T1
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
end
Month{1} % January
Month{12} % December
.
As always, my pleasure!
Just add only those variables —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
VN = T1.Properties.VariableNames;
[Y,M,D] = ymd(T1.date_time);
DN = day(T1.date_time,'dayofyear');
T1 = addvars(T1,M,Y,'AFter','date_time');
T1.Properties.VariableNames = {VN{1},'M','Y',VN{2}}; % Necessary, Since They Show Up As 'Var2' etc. Otherwise
T1
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
end
Month{1} % January
Month{12} % December
.
I assume you mean and empty column.
Probably the easiest way would be to create an empty table and concatenate it —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
VN = T1.Properties.VariableNames;
[Y,M,D] = ymd(T1.date_time);
DN = day(T1.date_time,'dayofyear');
T1 = addvars(T1,M,Y,'AFter','date_time');
T1.Properties.VariableNames = {VN{1},'M','Y',VN{2}}; % Necessary, Since They Show Up As 'Var2' etc. Otherwise
T1
NewVariable = table('Size',[size(T1,1),1], 'VariableTypes',{'double'}, 'VariableNames',{'NewVariable'});
T1 = [T1 NewVariable]
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
end
Month{1} % January
Month{12} % December
.
As always, my pleasure!
The easiest way to find out is to do the experiment, adding:
Month{k} = [Month{k}; []];
That will not throw an error, however it does not affect the result, and neither does:
Month{k} = [Month{k}; cell(1,size(Month{k},2))];
while this works:
Month{k} = [Month{k}; table(NaT,NaN,NaN,NaN, 'VariableNames',Month{k}.Properties.VariableNames)];
however this (and analogues of it) do not:
Month{k} = [Month{k}; table({},[],[],[], 'VariableNames',Month{k}.Properties.VariableNames)];
So it appears not to be possible.
I will leave this up here so you can experiment with it —
date_time = (datetime('01-Jan-2023') : caldays(1) : datetime('31-Dec-2023')).';
Temperature = randn(numel(date_time), 1);
T1 = [table(date_time) array2table(Temperature)]
VN = T1.Properties.VariableNames;
[Y,M,D] = ymd(T1.date_time);
DN = day(T1.date_time,'dayofyear');
T1 = addvars(T1,M,Y,'AFter','date_time');
T1.Properties.VariableNames = {VN{1},'M','Y',VN{2}}; % Necessary, Since They Show Up As 'Var2' etc. Otherwise
T1
for k = 1:12
Lv = month(T1.date_time) == k;
M = T1(Lv,:);
[~,idx] = maxk(M{:,2},5); % Index Returns 5 Highest Values Of First Variable
Month{k} = M(idx,:);
% Q = cell(1,size(Month{k},2))
Month{k} = [Month{k}; table({},[],[],[], 'VariableNames',Month{k}.Properties.VariableNames)];
% Month{k} = [Month{k}; table(NaT,NaN,NaN,NaN, 'VariableNames',Month{k}.Properties.VariableNames)];
end
Month{1} % January
Month{12} % December
.
その他の回答 (0 件)
This question is locked.
カテゴリ
ヘルプ センター および File Exchange で Dates and Time についてさらに検索
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
