Data availability analyse (Visual )

Hi gugys. I have a dataset:

K=65x1511 % where 65 is the number of houses and 1511 is the number of measurements.

What i want is to check the data availability of the measurements by plotting a figure like below, where the red pixels mean that the data is available while white mean that it is not available. But since i dont know which function is used to create such a data i'm totally lost.

Plz help.

[Geoff Hayes]

Hi Morteza,

So your data set is 65x1511 with the rows corresponding to a certain household (so 65 houses in total) and the columns correspond to various measurements for each household. But what do these measurements represent? Time values?

From your figure, the y-axis would correspond to the rows in your data set, with the x-axis being a period of time. Does that mean that for any row in your data set, the measurements correspond to fixed time periods (months?days?). And if so, would a particular value for that time period be either valid (and so red in figure) or invalid (and so white)? How do you determine the validity of a measurement?

Geoff

[per isakson]

See documentation on

    image
    imagesc
    pcolor

[Morteza]

Maybe i didn't explained myself well. Sorry. The measurments represent customers consumption per kwh. Actually I have:

K=24x63x65

where:

24= number of hours

63= number of days

65= number of customers

Thanks.

[per isakson]

• What does "K=24x63x65" mean? Is K a 3D numerical array?
• the picture shows a 2D array with the two values: "white" and "red"

[Morteza]

Yes, it is 3D numerical array

[Geoff Hayes]

So what does each dimension represent?

[Morteza]

it represent:

hours x days x customers

[Geoff Hayes]

So how does that information translate to the above image? The third dimension is customers and so corresponds to the y-axis, but how would you use the hours and days info for the x-axis?

So we have 63 days worth of information, with usage over a 24 hour period for each of those 63 days (for each of the 65 customers). So for one day, do you draw a red line (for customer i) for each hour that his/her energy consumption exceeds some pre-defined threshold? And leave as a white line otherwise?

[per isakson]

Could the information of K be represented by a 2D (24x63) array

    Customer

where

    Customer( iiHour, iiDay )

holds the number of customer for hour==iiHour and day==iiDay ???

Answer 1

[Star Strider]

I am assuming that if data are not available over all Hours and all Days, those data are set to zero. (If something else is the indicator, you will have to make appropriate changes to the code.) Briefly, I created a matrix of random integers that included zero, the took the product across either Hours or Days (since you did not specify what you wanted), so the data were available across all Houses. Any zero over Days or Hours for a particular House would then produce a zero for that House, indicating that at least one datum was missing. I then plotted them.

K = randi(30, 24, 63, 65)-1;           % Create Data

HouseXHrs = squeeze(prod(K,2));         % Product across ‘Days’
HouseXHrs(HouseXHrs >  0) = 10;
HouseXHrs(HouseXHrs <= 0) = 60;
HouseXDay = squeeze(prod(K,1));         % Product across ‘Hours’ 
HouseXDay(HouseXDay >  0) = 10;
HouseXDay(HouseXDay <= 0) = 60;

figure(1)
pcolor(HouseXHrs)
xlabel('Houses')
ylabel('Hours')
axis equal
colormap(jet)

figure(2)
pcolor(HouseXDay)
xlabel('Houses')
ylabel('Days')
axis equal
colormap(jet)

This produces for figure(2):

It is not exactly as you posted, but it is as close as I can get with MATLAB graphics.

[Morteza]

Thanks for the answer STAR.

But one problem. Costumer 8 has no data, since it is zero all the way up, which should be one color on the graph.

But this ain't seem like.

Thanks.

[Star Strider]

My data here are random integers. I don’t have your data, so I have no idea what your data look like when plotted this way. I simply provided you with a way to plot them yourself, using MATLAB graphics.

Insert your data for variable K and go from there.