indexing with isnan in multidimensional arrays
15 次查看(过去 30 天)
显示 更早的评论
I have this code:
load lon_nonan_reshape_use;
idx = ~isnan(lon_nonan_reshape_use);
lon_nonan_adj = lon_nonan_reshape_use(idx);
In the above example, lon_nonan_reshape_use and idx both are of size n*2 so I expect lon_nonan_adj also to be of n*2. MATLAB gives the results in n*1 dimension because lon_nonan_adj(:, 1) is not equal to lon_nonan_adj(:, 2) which is sensible. But I still want to get lon_nonan_adj as a n*2 matrix with NaNs retained in the place where there were NaNs previously. Could you please help?
4 个评论
DGM
2021-8-12
"But I still want to get lon_nonan_adj as a n*2 matrix with NaNs retained in the place where there were NaNs previously."
If you take matrix A, remove all the NaNs and then fill all the empty spots left over with NaN such that the size is preserved, then you're back at A. I don't see how the operation isn't superfluous.
采纳的回答
Walter Roberson
2021-8-12
load lon_nonan_reshape_use;
idx = ~all(isnan(lon_nonan_reshape_use),2);
lon_nonan_adj = lon_nonan_reshape_use(idx,:);
This retains rows provided there is at least one non-nan in the row.
1 个评论
Walter Roberson
2021-8-13
If you want to remove all rows with NaN anywhere in the row, then
rmmissing(lon_nonan_reshape_use)
更多回答(1 个)
Chunru
2021-8-12
% A small matrix with nans
a=randn(6, 3);
a([2 11 13])=nan
% idx
idx = ~isnan(a)
a(idx)
So a(idx) is a colum matrix based on the original a by removing all nans and arranged in a column vector.
If you want to remove all rows with nans (it looks so for your problem), then you can do the following
idx1 = all(~isnan(a), 2)
a(idx1, :)
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Matrices and Arrays 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!