Sorry because this topic recieve question but it seems that none respond to my situation…
I want to be able to use a cascading select from external files: the 4 administrative levels of 4 countries… you imagine there is quite a few lines!
Yyeeeaaaahhh i managed to make it work!
Had to leave, will upload the document later.
I used 5 levels
Type:
select_one_from_file csv_country.csv
select_one_from_file csv_region.csv
select_one_from_file csv_district.csv
select_one_from_file csv_community.csv
select_one_from_file csv_healthpost.csv
Name
pays
reg
dis
comm
stru
Choice-level
pays -> (Nothing for country)
reg -> pays=${pays}
dis -> reg=${reg}
comm -> dis=${dis}
stru -> comm=${comm}
With therefore 5 csv preloaded in UTF8 with first cell:
list_name,name,label,pays,reg,dis,comm
I used the column as choice filter as usual, only and i don’t know why i had to put only one list_name per level, therefore one csv per lecel. Not a big deal i upload the 5 csv not just one but its the same
Quite happy!
I’ll upload the files on my way back i wrote on my phone so i may have put a - instead of a _
I have to deploy a survey in 4 countries and locate until a 5th administrative level:
Country, region, district, community, health post
The cascading select through choice-filter and the normal choice page in cascade doesnt work due to the amount of rows to process.
I have been looking around and found different options but none seem to work with KoboCollect o OdkCollect:
Search in appearance works but doesn’t allow to filter. The option of the search function like ‘startwith’ or ‘endwith’, ‘match’ or even better column-to-filter would do the trick but do they work? It doesn’t seem to me that they work
Select_one_external is not aupported by Kobo apparently
Pulldata allow for pulling a value in a calculate type, not displaying choices of answers in a select_one as far as i understood
Select_one_from_file works when having one pre-loaded csv per level of cascade (5in my case) but only through enketo, not through the mobile apps
Please can someone helps me by telling me either to forget about it and just build as many survey as needed for the app to work or by giving me some examples of success or tricks?
Hi @Quentin,
I would like to recommend a different structural approach to sorting out this issue.
I assume that your region levels are manageable. So in this case you would need to have a region based skip which would allow you to divide the subsequent divisions into different questions which can be routed with a skip routine
By tricking the app that each list is different, you avoid the processing of a large list. I have done so by ensuring that the listing for each region is different. So you do not need external select but keep all the responses in the choices sheet…
You will create your cascade starting from the District not from the region, see example below.
Change your regions, districts, community and health posts appropriately. This should work out your issue. Please note if the Regions are still too big, think about the same logic but start the cascade at the Community.
Thanks Stephanealoo, i will try doing this
I’ll think about creating a vba code that would allow to do so automatically if i manage to find time i’ll share it over here
However, the great thing about external data is that these data can be updated while the form is deployed… i never tried but i think we cannot update a form while its deployed even if its just at the choice page level right?
If its true this solution is definitly not optimal because names of villages and health center change through time in such a big scale.
Hi @Quentin,
Hope you are doing well. I tried your approach but am getting the error message:
Unable to deploy form:
The same instance id will be generated for different external instance source URIs. Please check the form. Instance name: ‘project’, Existing type: ‘file’, Existing URI: ‘jr://file-csv/project.csv’, Duplicate type: ‘choice’, Duplicate URI: ‘None’, Duplicate context: ‘survey’.
Could you share your csv or a template with the first lines so we can check if there is an issue there?
Be sure to follow the example given with the fruits here to use the select_one_from_file
@Quentin Hi, thanks for the prompt response.
Sharing a snapshot of the first few lines of the first csv form (districts)
Sharing a snapshot of the first few lines of the first csv form (projects) - this is where i need the cascade. A project is a subset of district and must be filtered based on selection of a district
Hi @Quentin Could you help?
We are doing an urgent survey as a part of the state governments initiative for getting a better grip on COVID
Would really appreciate it if you show me how to get the cascading select with uploading external csv files, as described in the messages above
Sorry you know as things go in this sector we all are running late in our urgent tasks
I still don’t really see the csv you are refering to that should be in csv format UTF8 separated with coma that you call on you select_one_from_file function…
However have you tried doing it the simple way with your cascading options directly in the choice page with filtered by choice:filter columns?
In the survey project where you need your cascade, you have 2 questions one is select_one district named district
the other other one select_one project with as a value in column choice_filter igual to district = ${district}
and in your choice page you have under list name the two lists project and district with the slight difference that there is a third column called district where the value for each project should be the district
Always of great help.
Unfortunatly i still didn’t manage to make it work… for the moment we have splitted surveys by regions and aggregate de data after but i don’t really like this option.
I tried the files from the post in ODK forum and i didn’t succeed in making them work (without changing anything)…
For the moment i still recomand simple cascading questions and splitting surveys when it gets too heavy
Hi @Quentin
I am not promising that I will work on this immediately, could you however send me the form in the XLS version and the list as a separate excel. I will then try and look at it to see which scenarios would work. Kindly do so on a private message if you have sensitive information on the survey.
Note: Currently there are only 2 levels with around 500 rows. However the form i need to build has 100,000 rows and 4 levels, with the top level being a district, and there are 34 districts in the state.
Splitting it into 34 forms wouldnt make sense either because i would have around 3000 rows in each form.
Pulling from an external file makes the most sense, but i am unable to do so
Looking forward to your guidance here, and sorry for the trouble
Varun
Can I check if the problem remains unsolved? At CartONG we are providing an IM helpline for projects which are using data to assist the response to the coronavirus pandemic. https://www.cartong.org/cartong-im-covid-19-help-center
4 levels and CSV files not a problem. 100,000 rows we may need to do a little testing on to identify the exact performance limit and advise on the best way to manage that aspect. If this issue is still blocking you, do please let me know as soon as possible and hopefully we can help you get your form working.
Happy to help. Would you mind filling out this short form and I will contact you directly to your email address? No need to repeat the details of your technical problem as its already well understood so you can just write “already explained” if there’s any questions about that. If not working for an NGO you can just put the name of your employer/organisation, its just for our reporting needs and then I can contact you directly and see how we can move forward.
Hi, just wanted to share back with the community the relevant checks to do when using select_one_external and a single CSV file rather than multiple files.
With select_one_external the CSV file must be named “itemsets.csv”, other names are not possible
For reasons that I dont fully understand I have found that sometimes a form like this will be rejected if Labels, Hints etc do not all specify a language (eg. Label::English), even if there is only one language. Adding this helped avoid a server500 error.
When there are multiple cascading layers the top layer goes in the choices tab of your xlsform and all other layers are included in the CSV file.
System does not seem to have significant performance issues even with 100,000 rows
@Quentin, @Varun, kindly please be informed that you could manage your survey project through select_one_external question type as discussed in our previous post:
