Extraction, Data modeling and Analysis in different Scripts (Apps)

When implementing QlikView projects, I proceed with the following steps:

- Extraction of the source data and storing them in QVD format
- Transformation and implementing the data model
- Binary load the data model into the analysis application and implementing the user interface

For each of the steps above, I have different folders, well organized and I am able to easily implement step by step the project. Especially, data validation and the understanding of the input data and output is given, in order to validate the transformations and check the content of the tables for a source data analysis.

In Qlik Sense, we don't have the physical folders anymore, therefore how can we implement these steps in Qlik Sense?

First of all, the physical folders are still available, but not directly seen, because we implement in the Qlik Sense repository environment. However, this provides us with the ability that even we implement remotely on a web browser, the scripts are stored on the Qlik Sense server. Therefore any new folder data connection would actually show us the Qlik Sense server structure. And by this, I am able to store the tables in QVD files. So here is how I implement the steps above in Qlik Sense:

- Define the data connections for database and the QVD folder (e.g. D:\QlikSense\Extraktion\) and create the extraction script
- Extract data as usual and store the table with the store command. I have created the required folders for storing on the Qlik Sense server

store Statistics into [lib://Extraction/Table1.qvd] (qvd);
drop Table1 Statistics;

- Create the data model script and implement the data model, by using the "Extraction" connection to read in the QVDs
- In order to binary load the data model, I create a new folder data connection to the Qlik Sense application folder (C:\ProgramData\Qlik\Sense\Apps\). The files stored here are not the names we set when we created the scripts. The file names are stored in form "1AFB6709-6C8E-4BCF-BD17-306BAD377A9C", so I have to check for the file on the server once (based on the modified data), which script is my data model. There might be another way to find out the file name of my data model, but I could't figure it out. Then I binary load the data model into my analysis application. Even the format of the file is not QVW, you will be able to binary load the table, by selecting "All datatypes" in the datatyp drop down
binary [lib://Sense_Apps/1AFB6709-6C8E-4BCF-BD17-306BAD377A9C];

At beginning, it seems to be not as convenient as it was with QlikView, however, if I can figure out to find the data model name in order to binary load into the application, the proceeding with extracting, data modeling and binary load can easily implemented as having the physical folders underneath.



blog comments powered by Disqus