Basic rules for creating a perfect data model - multiple data sources

[Anonymous]

I require some help and guidance - best with some examples if possible- that would really help

Basically what are the rules for creating a data model which is good enough to start development.

I have a requirement to connect to multiple data sources such as Excels, CSVs, Databases etc.

After the required tables are pulled in, how do we identify the joins between these tables.

please send me your advice help guidance.

thank you so much in advance.

[Anil_Babu_Samineni]

Looks like similar Design & Development Standards Best practices required

FYI - Best Practices for Data Modelling

[Anonymous]

Thank you for the link. Will go through it.

How do we identify the primary keys in multiple tables if we are having multiple tables from multiple data sources.

Basically rules to create a data model to kick start a project. Please share some tips in data model creation.

[Anil_Babu_Samineni]

If you look data model after loading all tables, There is one common field called "Primary Key". That means column name and column sort order for each primary key of filter the Schemaname, Tablename..

Off course, If data model is not good that means you will be frustrate on data points in UI. So, Initially we need to create perfect data model which never explanation come accuracy of 100%. So you need to read minimum 50 blogs related data model then only somewhat we are walking through nice data model and creating is good until unless data is mismatching. Will this helps to you?

[Anonymous]

Require a sample example of a good data model which explains these primary key concepts with the joins. See if you can send me a sample and explain more.

[Or]

You're asking two entirely different questions, if I read your original post correctly:

1) How do we determine the relationships between data sources?

Answer: This has nothing to do with QlikView. This is part of the data preparation work that you should be doing with the people who understand the data sources. At the end of the day, QlikView will create a schema based on fields that share the same name, so it's your job to rename fields correctly. Your input as a QlikView specialist is needed primarily in situations that don't have a natural relationship, such as master calendar situations or circular loops that need to be broken down with relationship tables.

2) What is the best-practice schema for different situations?

Answer: This is too broad a question to really answer. It depends on too many factors. Two good resources have been linked in a previous post, and additional resources can be found via Google search or possibly in the Qlik Continuous Classroom (which is available free this coming Wednesday through Friday).

Personally, I don't worry too much about setting up the "perfect" schema. It needs to work, and it needs to be human-readable enough that I can maintain it moving forward, but that's pretty much it. Unlike traditional database-level schema, there's usually no significant advantage to using a star / snowflake schema in Qlik other than readability.

[rwunderlich]

Or, I would give you double "likes" if this system allowed it, especially for your last paragraph about the "perfect" schema.

The natural association(linkage) based on common fieldnames (mentioned in your paragraph 1) is sometimes confusing and surprising to newbies, but it's where the important work is and makes the associative magic happen.

-Rob