Find minimum values based on unique number

I have a dataset of temperatures collected over multiple depth profiles (1-147 profiles). The data is all listed in one long table, not by each profile (attached).

Each profile has a different temperature minimum, and I want to find this minimum for each profile, and colour all of the temperatures above this in grey in a figure (essentially discarding them).

1. Evidently I'm going about this the wrong way as my output (T_min) is all the same number (see code below).
2. Once I have the T_min for each profile, when I do a scatter plot, how can I colour each dot less than the T_min - for that particular profile - grey?

Thanks in advance - sorry if this isn't very clear.

j=1;
for i=1:length(dives)
    T_min(j) = nanmin(SG579_FINAL_table_MF.Cons_temp(dives));
    j=j+1;
end

Answer 1

[Mathieu NOE]

hello

the provided mat file contains 492 different profiles (and not 147)

my suggestion so far :

load('matlab.mat')

prof = Prof_temp.Profile_num;
prof_unic = unique(prof);

for k=1:length(prof_unic)
    ind = (prof == prof_unic(k));
    T_min(k) = min(Prof_temp.Temp(ind),[],'omitnan');
end

plot(prof_unic,T_min);
xlabel('Profile #')
ylabel('Average temperature')

[DD_2023]

Apologies, I tweaked the data before I shared it and forgot to update my text.

Thank you so much, your code has highlights the minimum values in each profile, exactly as I wanted!

I was now hoping to plot a scatter plot of the data against another variable (called S), but colour the points where the points less than the T min for that profile are coloured differently.

[Mathieu NOE]

and can you provide this S data ?

[DD_2023]

Of course, and as you may have seen below I have misexplained what I actually want to do.

This is my aim:

T_min for each profile will be located at a certain depth (called Pres, and the T_min Pres will differ between profiles). In the scatter plot of T and S, I wanted to colour any dots shallower than the T_min Pres (differing between profiles) grey.

[Mathieu NOE]

if I have understood correctly maybe then this is it :

load('matlab.mat')
load('Pres.mat')
load('S.mat')

prof = Prof_temp.Profile_num;
prof_unic = unique(prof);

indgrey = [];

for k=1:length(prof_unic)
    ind = (prof == prof_unic(k));
    [T_min(k),indm] = min(Prof_temp.Temp(ind),[],'omitnan');
    Pres_of_Tmin(k) = Pres(indm);
    % indices of data to put in grey
    indgrey = [indgrey; find(Pres(ind)<Pres_of_Tmin(k))]; % find for each profile the data for which Pres is lower than Pres_of_Tmin
                                                          % and concatenate those indexes
    
end


indnongrey = (1:numel(prof));
indnongrey(indgrey) = [];

scatter(S(indnongrey),Pres(indnongrey),5); % "non grey" elements
hold on 
grayColor = [.7 .7 .7];
scatter(S(indgrey),Pres(indgrey),5,grayColor); % "grey" elements
hold off
xlabel('S')
ylabel('Temp')

[DD_2023]

Thanks so much, this code works with my data, but looks like it is maybe only applying to one profile at a time rather than finding the depth minimum T for every profile and greying out all data shallower than that?

Apologies if it isn't doing that - I just would have expected more points to be grey

[Mathieu NOE]

the code - hopefully - should do what i put in the comment (and what I understood from your query but maybe I'm wrong)

% find for each profile the data for which Pres is lower than Pres_of_Tmin

% and concatenate those indexes

[DD_2023]

I don't think its far off, but I would expect most data at a pressure less than somewhere between 50 - 200 m to be grey. But I don't want to take up any more of your time, I'm sure I can fiddle with what you've done for me so far, it's a huge help - thank you!

[Mathieu NOE]

I have now a slight doubt when you say : one profile at a time rather than finding the depth minimum T for every profile and greying out all data shallower than that?

does "all data" mean data of each profile or all together ?

my code above was supposing greying within each profile

if it's the second option, the new code would be - hopefully again without error :

load('matlab.mat')
load('Pres.mat')
load('S.mat')

prof = Prof_temp.Profile_num;
prof_unic = unique(prof);

indgrey = [];

for k=1:length(prof_unic)
    ind = (prof == prof_unic(k));
    [T_min(k),indm] = min(Prof_temp.Temp(ind),[],'omitnan');
    Pres_of_Tmin(k) = Pres(indm);
    % indices of data to put in grey
    indgrey = [indgrey; find(Pres<Pres_of_Tmin(k))]; % find the entire data which Pres is lower than Pres_of_Tmin
                                                          % and concatenate those indexes
    
end

indgrey = unique(indgrey);

indnongrey = (1:numel(prof));
indnongrey(indgrey) = [];

scatter(S(indnongrey),Pres(indnongrey),5); % "non grey" elements
hold on 
grayColor = [.7 .7 .7];
scatter(S(indgrey),Pres(indgrey),5,grayColor); % "grey" elements
hold off
xlabel('S')
ylabel('Temp')

[DD_2023]

Thank you! This looks more like what I was expecting. Apologies for my lack of clarity :)

[Mathieu NOE]

ok - glad we finally converged to the solution !! :)

Answer 2

[Stephen23]

P = load('matlab.mat').Prof_temp

P = 460116x2 table
    Profile_num     Temp  
    ___________    _______

         1         0.73595
         1         0.73338
         2         0.73372
         2         0.72239
         2         0.73413
         2          0.7322
         2         0.73786
         2         0.73755
         2         0.73594
         2         0.74415
         2          0.7336
         2         0.72792
         2         0.74704
         2         0.75886
         2         0.75926
         2          0.7642

S = groupsummary(P,"Profile_num","min","Temp")

S = 492x3 table
    Profile_num    GroupCount     min_Temp 
    ___________    __________    __________

         1              2           0.73338
         2            144           0.65173
         3              2           0.73312
         4            144           0.69801
         5            239           0.36951
         6             78           0.35654
         7             79           0.70014
         8            160           0.66216
         9            156          0.013414
        10            205          0.013662
        11            442        0.00084435
        12            626          0.028878
        13            705          0.025135
        14            959          0.015209
        15            683          0.026312
        16            915         -0.058026

plot(S.Profile_num,S.min_Temp)

[DD_2023]

Thank you so much!

I was now hoping to plot a scatter plot of the data against another variable (called S), but colour the points where the points less than the T min for that associated profile number are coloured differently. (Sorry, I know this is confusing wording).

[Stephen23]

" colour the points where the points less than the T min for that associated profile number are coloured differently."

I guess you mean where the S-values are less thatn T-min (because of course no Temp values should be less than T-min.

P = load('matlab.mat').Prof_temp

P = 460116x2 table
    Profile_num     Temp  
    ___________    _______

         1         0.73595
         1         0.73338
         2         0.73372
         2         0.72239
         2         0.73413
         2          0.7322
         2         0.73786
         2         0.73755
         2         0.73594
         2         0.74415
         2          0.7336
         2         0.72792
         2         0.74704
         2         0.75886
         2         0.75926
         2          0.7642

S = randn(size(P.Temp))

S = 460116x1

-1.9744 -0.7961 -0.0034 -0.2607 -0.4320 -1.1208 0.1508 0.5829 0.5262 -1.6224

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

V = grouptransform(P.Temp,P.Profile_num,@min);
X = S<V;
scatter(S(X),P.Temp(X),23,[1,1,1]./2)
hold on
scatter(S(~X),P.Temp(~X),23)
xlabel('S')
ylabel('Temp')

[DD_2023]

Sorry you make an excellent point and have made me realise I've completely messed up my explanation.

T_min for each profile will be located at a certain depth (called Pres, and the T_min Pres will differ between profiles). In the scatter plot of T and S, I wanted to colour any dots shallower than the T_min Pres (differing between profiles) grey.

[DD_2023]

Here are the data for Pres and S