Downsampling array data from counts per minute to counts per hour

I have an evenly sampled array of data, say chickensPerMinute.

I want to downsample this by a factor of 60 to return chickensPerHour.

Is there a MATLAB function to do this, akin to the downsample function which returns mean values for each downsampling window.

Answer 1

[Star Strider]

That depends on what you want to do.

I would use a timetable and the retime function —

Time = datetime(2004,10,23,17,00,00)+minutes(0:299).';
Chicken_Rate = randi(9,numel(Time),1);

TT_Chickens = timetable(Time,Chicken_Rate)

TT_Chickens = 300x1 timetable
            Time            Chicken_Rate
    ____________________    ____________

    23-Oct-2004 17:00:00         5      
    23-Oct-2004 17:01:00         7      
    23-Oct-2004 17:02:00         7      
    23-Oct-2004 17:03:00         2      
    23-Oct-2004 17:04:00         5      
    23-Oct-2004 17:05:00         6      
    23-Oct-2004 17:06:00         4      
    23-Oct-2004 17:07:00         2      
    23-Oct-2004 17:08:00         5      
    23-Oct-2004 17:09:00         4      
    23-Oct-2004 17:10:00         8      
    23-Oct-2004 17:11:00         5      
    23-Oct-2004 17:12:00         3      
    23-Oct-2004 17:13:00         9      
    23-Oct-2004 17:14:00         7      
    23-Oct-2004 17:15:00         4      

TT_Chickens_Hourly = retime(TT_Chickens,'hourly','sum')

TT_Chickens_Hourly = 5x1 timetable
            Time            Chicken_Rate
    ____________________    ____________

    23-Oct-2004 17:00:00        326     
    23-Oct-2004 18:00:00        296     
    23-Oct-2004 19:00:00        266     
    23-Oct-2004 20:00:00        281     
    23-Oct-2004 21:00:00        303     

.

[dormant]

Many thanks for introducing me to a new way of MATLAB. timetable looks as if it could also be useful in other scripts,

[Star Strider]

As always, my pleasure!

The retime function is extremely useful for this and other operations. Another is the synchronize function.

Answer 2

[ScottB]

if A represents chickensPerMinute then

B = A(1:60:end); should downsample to chickensPerHour just by indexing.

[dormant]

That doesn't work:

>> A = [ ones(1,90) zeros(1,120) 2*ones(1,150) ];
>> B = A(1:60:end)

B =

     1     1     0     0     2     2

Answer 3

[Steven Lord]

Another potential approach would be to use the groupsummary function.

[dormant]

Thanks, I've used groupsummary before, but forgotten about it.

Answer 4

[Harsh Sharma]

Hi dormant,

There can be multiple ways to downsample your data depending on the initial data type. But if your input is just an evenly sampled array you can use the “reshape” and “mean” functions in MATLAB to achieve your desired result. Below is an example code that illustrates how to do so:

% An array containing chicken values for 120 minutes
chickensPerMinute = randi([1 10],1,120)

chickensPerMinute = 1×120

1 1 8 1 2 6 3 8 3 5 5 6 10 9 5 5 8 4 10 6 2 8 6 1 6 7 3 1 8 10

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

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

% each column in reshaped array corresponds to chickens in the respective hour
reshapedArr = reshape(chickensPerMinute,60,[])

reshapedArr = 60×2

1 1 1 9 8 9 1 9 2 5 6 1 3 1 8 9 3 8 5 8

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

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

% taking mean in column direction gives the avg chickens per hour
chickensPerHour = mean(reshapedArr,1)

chickensPerHour = 1×2

5.5000 5.2000

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

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

You can look at the following documentation link to understand how “reshape” function works - https://www.mathworks.com/help/matlab/ref/double.reshape.html