chrislarsenqlik / SenseBoxPlot

3 stars 5 forks source link

Wrong Calculation in the Box Plot #3

Open dewaputrabs opened 8 years ago

dewaputrabs commented 8 years ago

Hi Chris,

I use your boxplot graph to create a report view and the things i want to analyse is the download speed per region.

i show all region in pie chart and give the download data in boxplot chart.

but some how the view is getting wrong. when i show all the region the result will not the same with when i filter only one region.

please find the attach for the sample, i hope you got what i mean.

allregiondownloadanalysis

westjavadownloadanalysis

thanks any way :)

chrislarsenqlik commented 7 years ago

Hi, You can see that it does change but the number of cities on the x axis it too much for the space designated and there is too much to do layoutwise to accommodate it using d3. Instead I will be switching to High Charts for the box plot. I will respond back here. Until then, please limit the number of dim values to be just the top 10 or something that works with your layout.

dewaputrabs commented 7 years ago

Hi Chris, i know its change, but what i mean is not city boxplot, my point is in the region boxplot part. you see i only have 8 regions. and the information is miss match. when i show all 8 regions and only one region.

thanks.

chrislarsenqlik commented 7 years ago

I am confused, sorry. In the screenshot i see you made a selection on West Java and in the Region box plot displays only that region for West Java. What happens when you cancel or clear the selection(s)?

dewaputrabs commented 7 years ago

Hi Chris,

you seen my first screenshot ?

its show all regions ( 8 regions there and west java is one of them with 4th graph ) with value Q1= 92; Median = 151 ; Q3= 221

but in my second screenshot. i chose only one region ( west java region ) and the value is change.
as you can see when i choose only west java region Q1=269 ; Median = 586 ; Q3 = 1084.

and recently i found every time i load the dashboard the data view is some how keep changing :(

may be you should look again for the Quartal calculation.

jon-nagra commented 7 years ago

I found the same problem. I believe the package takes only the unique values for the second dimension. I uploaded a dataset that makes it clear the boxplot is not working properly (the median should be 1 for variable y).

chrislarsenqlik commented 7 years ago

Thanks I'll check this out

Haury commented 7 years ago

Hi Chris! First of all thank you for this extension. As mentioned in the previous posts. The Boxplot seems to have a Problem if there are entries with the same value in a table. I attached a screen showing my basetable, a table with the calculations (Fractile, Median) and your Boxplot. The Boxplots for A and C are correct as there are now duplicate values. B on the other hand is wrong.

It would be great if you could check on that again! Thanks in advance!

boxplot wrong values