Red lines at my data model

[#1]

Hi guys.

I've made a odbc connection and it worked well but suddenly these red lines showed up in my data model view.

What is this?

I've checked and there's no wrong model or wrong keys in this connection.

Thanks in advance.

[#2]

This application is "mine", I'm doing this projet for a client. Very soon my client will get the Qlik Serve so this app aren't going to be local anymore. I'm connecting to its ODBC and consuming some data to tests. I'm afraid that futher this app doesn't work anymore or have this kinda of "problem" again.

[#3]

If i rename and lost de connction again, how can i reconnect my data without use synthetic key? Sorry, I'm very new using Qlik Products. I think that its the problem: almost all my fields are keys.

[JonnyPoole]

You have to use alias to rename the fields otherwise it will look for the new name in the data source.

change:

load

     Date

from <>

to

load

     Date as NewDate

from <>

the links will form using the alias name which comes after the 'as'. Qlik will still look for the field 'Date' in your data source so that you shouldn't get connectivity/read errors.

[#5]

I've already done this Jonathan. My problem is, even changing the name didnt work. I need the relation with surveu_request, survey_queue, survey and survey_question. These 4 tables has only the field survey_id as connection. When I rename this relation/sync desapears. So in my visualizations the data seems to be lost.

I've tried everithing that you guys said but nothing is working and this circular reference stills here.

[reddy-s]

Hi Gabrielle,

If you rename the request id field to some thing else , you can get rid of the Circular reference now.

[JonnyPoole]

I *THINK*  it may serve you to join the survey answers/questions into 1 table and survey requests/votes tables into another.You can use inner/outer and right/left joins in the load script to do this. The link between the two wider resulting table would be a synthetic key of both survey_id,and question_id which you could replace with a bonafide link table later on as you fine tune this. 

But before doing this, is the relation between vote and request one to many or many to many ? joining one to many tables is usually ok but joining many to many tables usually requires a table concatenation in lieu of a join to ensure you don't get double counting.

[simondachstr · Accepted solution]

I would recommend to you to start from the beginning and adding the tables one by one. Following the conversation in this thread, you don't really understand the data model in my opinion. This is not your fault, however, in order to display accurate visualizations and aggregations, as well as extending the model with enhanced data, this will only work if you truly understand what's going on and why things are connected the way they are. Hence, you should start from the beginning one by one rebuilding the current model and in my opinion it should not take long. If you don't, you might end up creating a patched elephant which will be uncontrollable and do whatever it wants.

[#9]

Thanks for the advice. Its true that I'm lost and I'll do it all again to understanding what data model is and how it works. I'll pick all material that u guys gave to me and im going to analyse the docs and videos. Thank you so much guys. You're amazing!

Best regards to all you.

Gabrielle Souza