How about to separate two action filters to respective data source?
Action Filter 1.1:
Run On : SELECT
Source Sheets: Data Source 1
Target Sheets: Visualization 2
Target Filters: DataSource1.KeyField1 = DataSource1.KeyField1
Action Filter 1.2:
Run On : SELECT
Source Sheets: Data Source 2
Target Sheets: Visualization 3
Target Filters: DataSource1.KeyField1 = DataSource2.KeyField1
I tried your suggestion and few more possibilities. Data source 2 is
independent and visualization 3 is based on it. This does not seem to work.
I believe Action Filter "SELECT" only works in response to the UI Event. If
user performs UI operation on Visualization 1 then it will trigger refresh
of Visualization 2 because of Action Filter 1 BUT it is not triggering
Action Filter 2 because there's no
UI event on Visualization 2. Am I missing something?
Is there a workaround for this?
Thank you so much for your time. I really appreciate it.
On Wed, Oct 10, 2018 at 4:51 PM Shinichiro Murakami <
Want to have mocked up sample data for further investigations.
Here is the mocked up sample workbook and sample data. Here are some of my
There's no direct linking or blending between analysis vs. patients and
- Patients Visualization is able to update all visualizations.
- Features Visualization can not update analysis but all other
- PatientIds visualization is able to update all visualizations.
So why Features visualization is not able to update analysis but everything
else and what should be the solution to this?
On Wed, Oct 10, 2018 at 5:19 PM Shinichiro Murakami <
SampleData.zip 149.9 KB
To make action filter work across worksheets, two sheets need to have shared dimension.
In this case, you need to add Patient ID to Features View.
Then filter works with Patient ID as filter trigger.
Now, alternative problem is to eliminate the duplication on "Feature" viz.
If this is Okay, please mark my answer as correct to close the thread, not from inbox but from original post.
SampleSheet_SM_10.3.twbx 71.7 KB
I just realized that this will not be the correct solution. In features I have "Male" and "Female" grouping. When we pick First() = true then it will only take first record and my "Analysis" set will only show the first "Male" or first "Female" data instead of group by aggregate. So First() = True won't work in this situation BUT yes, Action filter does work using this.
Are you asking further help with another data set?
Shin, Can we continue the discussion further with the same dataset because
suggested solution is partially solving the problem. Yes this method is
able to trigger Action Filter setup between on two independent data sources
BUT this creates another issue with the aggregations for the dataset and
On Thu, Oct 11, 2018 at 12:14 PM Shinichiro Murakami <
Could you explain the New issue clearly with number example ?
I am not sure current sample is sufficient to replicate your issue or not.
In the sample dataset we have 6 female patients.
So if we click on "Female" in the dashboard within Features then we should see 6 patients in "Patients" Visual, 6 patient Ids in the patient Id vis and in Analysis vis we should see BoxWhisker of 6 patient features.
After using First() now we can't see all 6 but only the first patient Id and his related information.
So now, go back to very initial concept,
You said source 2 = Analysis is huge, then Analysis as starting data source (Left most)
I don't know your challenged this and I am proposing circular referencing idea or not, but this is the way to achieve the expectations.
Patient ID as key field, join two other sources.
SampleSheet_SM_10.3_v2.twbx 46.3 KB
This was my initial attempt. If my analysis data-set is small e.g. couple of thousands then it works but when it goes into millions then system takes a very long time. With each click I'm spending 40-50 minutes to get system response.
So the problem is 2 part:
Part 1 find Patient IDs by selecting any combination of values from different tables [This is working fine]
Part 2 Render analysis for Identified Patient IDs. [This is the issue]
I am taking Patient IDs as simpler example but in actual I have a Hierarchy of Study-> Experiment -> Patient -> Biopsy -> Image -> Image Features -> Analysis
It is quite difficult to investigate your real data source, but million rows does not seem like to big.
And if other two table is only adding attributes to big data source, it also should't consume such longtime.
I am not an expert of server side, but I recommend that you check different areas of long rendering.
I really appreciate your effort in resolving my issue. My challenge is a bit different. It is not just about joining millions of rows but being able to Multi-Select different values of different datatypes from different table columns to find related analysis data. Each data type visualization is different. I wish Sets were used across disjoint data sources. This could help me solve this issue. I will keep looking for solutions and I will update this post with the most satisfactory solution for benefit of others.
I can only create a combined set for set(s) created within one data source and I can create a filter based on this combined set. I can't create combined sets across disjointed data sources as far as my knowledge goes.