find the element location

I have returns marked in a column of 1*20million column. It has a specific value as -10, but I am not able to find out where it is.

I have tried using

find(returns(:,1)==-10)

but couldnt interpret the ouput. -

ans =

0×1 empty double column vector

When I check this matrix, its blank and does not even give any answer. Can someone guide on this?

I also wanted to plot these returns and histogram for this-

For plot-

plot(returns)

But I just get the x and the y axis. The reason could be i have 20 million rows and thus, its not visible in the graph. Can we show these values, in this graph, rather than just showing the x and y axis as it gives a vague picture.

For histogram-

nbins = 4000; %i have calculated the appropriate bins

hist(filteredreturns,nbins)

However, I am just getting everything at a point. I want to see the distribution and its should mostly be normally distributed or highly skewed towards right.

Answer 1

[Star Strider]

You have encountered ‘floating-point approximation error’. Most numbers are not exactly represented in floating-point precision (especially if they are the result of computations), so testing for them exactly is not usually productive. See the documentation on Floating-Point Numbers for an extended discussion.

Taking that into account, finding the indices of the ‘-10’ elements is relatively straightforward.

For example:

Returns = (0.5-rand(1, 1E+4))*30;                           % Test Vector
Lv = ismembertol(Returns, -10, 0.01);                       % Logical Vector Result
Indices = find(Lv);                                         % Index Positions

Use your vector instead of my test vector. Experiment with the tolerance value (here 0.01) to get the result you want.

See if plotting with a marker (instead of a line) will make your data more visible. I cannot offer you any advice with respect to the histogram. We do not have your data or know what result you want, so you will need to experiment.

Answer 2

[Guillaume]

If find returns empty, then we can safely say that exact -10 is not in that first column. Perhaps the number you're looking for is very close to 10 but not exact 10 (e.g. it could be -10.0000000000000017764). To find the nearest number to -10:

[~, loc] = min(abs(returns(:, 1) + 10)
diff_from_10 = returns(loc, 1) + 10

As for the rest, I don't understand what you're saying.

Note that hist has long been deprecated. histogram is the recommended function now.

[Harsh Rob]

Thanks Guillaume