Very Large Box and Whisker Datasets

[TheWitness]

If I want to create a Box and Whisker chart, but each dataset contains millions of rows, is there a way I can send the final data (median, max, min, +-3std deviation) to the chrart instead of the entire set for dataset?  I think sending a million rows for dataset, that rendering will be extreemly slow.

 

Please advise.

 

Larry Adams

[Haritha]

Hi Larry,

 

The Box and Whisker chart calculates the mean, median, upper and lower quartiles and the minimum and maximum numbers for a given set of data automatically as of now. 

 

Hence, you need to provide all the set elements of the dataset  to the chart.

 

Hope this helps.

Edited July 16, 2013 by Haritha

[TheWitness]

You have no idea how disappointed I am.  Since we are under support, is it possible to modify the source to allow us to modify the behavoir?  This chart is exactly what we need, but our record set's are just too large to be passing over a httpd connection.  Without being able to pass the calculated values directly, this chart becomes USELESS...  Sadly.  So, modification of source is what we would need to do....

[Haritha]

Hi,

 

It is difficult to obtain the values of dataset elements from mean, median, upper and lower quartiles and the minimum and maximum numbers and plot the chart accordingly, if calculated values are provided to the chart. Hence, this feature is not supported and you will have to provide all the datasets that you need to plot on the chart.

 

Hope this helps.

[TheWitness]

I will contact the sales organization to log a feature request.  It is not practical to pass millions of data points over httpd.

[Haritha]

Hi,

 

We are looking into the feasibility of your request and will get back to you soon.

[shamasis]

This is interesting!

 

If I understood you correctly, you have millions of points as such you would like to send pre calculated info in the set elements?

[TheWitness]

That is correct.  We will offline calculate the various data points (min, max, median, average, 1-quartile, 3-quartile, etc.) in a daily ETL process.  That way the UI can be very responsive to all subsequent user requests.  The data will go from tens of millions of records per day, to tens of thousands of pre-calculated records per day, and for the specific graphing function, dozens of records.

Edited July 23, 2013 by TheWitness

[shamasis]

Let me check feasibility and revert. It might not be simple upfront since the chart's purpose itself is to calculate from value and has been designed accordingly. Will revert.

 

Since you're already doing the calculation, I presume providing it as a comma separated five number summary in a specified order is a good way for data input?

 

Finally, can you check and let me know if your license has unobfuscated source code?

[Duplex]

I also have the need for passing a pre-calculated five-number summary to the box plot.  Honestly, I'm pretty amazed that this isn't an included feature.

Edited August 12, 2013 by Duplex

[Haritha]

Hi,

 

Welcome to FusionCharts forum 

 

The feature of passing pre-calculated values to Box and Whisker chart will be included in our next release. We shall update this thread when our next version releases.

 

Hope this helps.

[Cesar]

Hi..

 

excellent news, but when will be the next version? a tentative date?

 

I was checking the last release but I didn't see that feature yet, and we need passing pre-calculated values to Box and Whisker chart.

 

Thanks a lot.

[Sanjukta]

If I want to create a Box and Whisker chart, but each dataset contains millions of rows, is there a way I can send the final data (median, max, min, +-3std deviation) to the chrart instead of the entire set for dataset?  I think sending a million rows for dataset, that rendering will be extreemly slow.

 

Please advise.

 

Larry Adams

Hi Larry,

 

Currently this is not supported.

 

But we have already reported this and shall update you in case we come up with a solution to this,

 

Thank you for your continued patience. 

[Cesar]

Hi Haritha,

I was checking about new versión, its great, but with this one, can I provide all the set elements of the dataset to the chart?  avoid to calculate?

I need a simple, pls

 

Thank in advance

[shamasis]

Yeah! You can. We will update the docs soon!

 

Now q1, q3, min, max, median and md, sd, qd, mean can be given directly as attributes of <set> element.

 

If the value attribute is given in <set>, then the chart will calculate these values automatically.

[xuejiaye]

Yeah! You can. We will update the docs soon!

 

Now q1, q3, min, max, median and md, sd, qd, mean can be given directly as attributes of <set> element.

 

If the value attribute is given in <set>, then the chart will calculate these values automatically.

So great!

    Which version contains this? I check the docs ,But I can not find it!

    Would you give me a sample? 

   Thanks a lot!

[Swarnam]

Hey,

 

This feature is supported in our latest version[FusionCharts XT v3.4]. As mentioned earlier, we will update the docs soon.

 

Sample in the following JSFiddle link: http://jsfiddle.net/swarnam/vLzadr81/1/

 

Hope this helps.

[openmet]

Hi;

I'm using release v3.13.1-src1, and I need to use Box and Whisker charts with tens of thousands of values, so sending all of the values to the browser is not going to work efficiently.

I understand from this thread that this chart can also be displayed if I calculate the median, q1, q3 etc. but I cannot find it in the documentation.

Please can you confirm whether this is supported and provide some documentation on how to do it?

Thanks for your support,

       Jordi

[Akash Biswas]

Hi Jordi,

 

The Box and Whisker chart of FusionCharts is a statistical chart that shows a frequency distribution drawing a statistical conclusion for the given data using the five number summary principle. It provides the feature to show mean, median, upper and lower quartiles, and the minimum and maximum numbers for a given set of data. It also calculates and displays the mean deviation, standard deviation, and the quartile deviation for the given set of data.

Please check the below documentation links for reference :

Features of Box and Whisker chart : https://www.fusioncharts.com/dev/chart-guide/standard-charts/box-and-whisker-chart#features-1

Five number principle : https://www.fusioncharts.com/dev/chart-guide/standard-charts/box-and-whisker-chart#fivenumber-summary-principle-2

Sample fiddle : https://jsfiddle.net/fusioncharts/qLvZF/

 

Box and Whisker chart is a statistical chart that is not developed to visualize huge dataset. FusionCharts provides separate chart types that supports huge dataset, please check the documentation link for reference : https://www.fusioncharts.com/dev/chart-guide/standard-charts/zoom-line-charts

Please note : These chart types supporting huge dataset does not support statistical chart functionalities like Box and Whisker chart.

 

Thanks,

Akash.

[openmet]

4 hours ago, Akash Biswas said:

Hi Jordi,

 

The Box and Whisker chart of FusionCharts is a statistical chart that shows a frequency distribution drawing a statistical conclusion for the given data using the five number summary principle. It provides the feature to show mean, median, upper and lower quartiles, and the minimum and maximum numbers for a given set of data. It also calculates and displays the mean deviation, standard deviation, and the quartile deviation for the given set of data.

Please check the below documentation links for reference :

Features of Box and Whisker chart : https://www.fusioncharts.com/dev/chart-guide/standard-charts/box-and-whisker-chart#features-1

Five number principle : https://www.fusioncharts.com/dev/chart-guide/standard-charts/box-and-whisker-chart#fivenumber-summary-principle-2

Sample fiddle : https://jsfiddle.net/fusioncharts/qLvZF/

 

Box and Whisker chart is a statistical chart that is not developed to visualize huge dataset. FusionCharts provides separate chart types that supports huge dataset, please check the documentation link for reference : https://www.fusioncharts.com/dev/chart-guide/standard-charts/zoom-line-charts

Please note : These chart types supporting huge dataset does not support statistical chart functionalities like Box and Whisker chart.

 

Thanks,

Akash.

Hi Akash;

Thanks for reply. However I don't understand the rational for not supporting this type of charts for large datasets.

A Box and Whisker chart is a chart that is used to describe a set of numbers. It describes a dataset regardless of its size. I do not understand why my end user should only see this chart when it comes from a small amount of data and should not see it if it comes from a huge dataset of numbers. Even more, statistical information makes more sense for large amounts of data.

FusionCharts is a charting library, not a statistics library, so I should be able to display the best charts to describe the data that I want to communicate to the end user. On top of this, the way we manage data is by applying weights to the different numbers of the dataset (not all data points have the same weight), so in this case letting the chart to calculate the statistics is not going to work either.

I really think that as a charting library FusionCharts should seriously consider to provide a way to display Box and Whisker charts for large datasets. If this means that key metrics (median, q1, q3, etc.) should be calculated in advance and then given to the library to plot the intended chart, this is fine.

I bought the PowerCharts license mainly for this type of chart. As our application manages large amounts of data I need your support to find a workaround that allows us to display Box and Whisker charts for large datasets.

Please provide some assistance and help.

Best regards,

           Jordi