If you are following the "Exercises - Basic", you can skip this exercise.
In this exercise you will use the data from the core application and build a Calculation View to analyze the Twitter sentiments. Therefore, the Tweets CDS table as well as the provided HANA Full-Text Index output (containing the sentiments) are used and joined together.
The planned duration for this exercise is 20 minutes. You can find a recording of the exercise here.
Run the following steps to complete the exercise:
Right-click on the src folder (ain365-template > db > src) and select New > Folder.
Enter models as Folder Name and click OK.
Right-click on the newly created models folder and select New > Calculation View.
Enter tweetSentiments as Name, Tweet Sentiments as Label and click Create.
NOTE
We need to add three parameters with multiple values to the calculation view and map it with the view from the synonym. To simplify this step, we insert the parameters directly to the XML file.
Right-click on the newly created calculation view "**tweetSentiments.hdbcalculationview**" and select **Open Code Editor**.
Mark the text <localVariables/> on line 4 and replace it with the snippet copied from the documentation repository.
Don't forget to save the XML file.
The XML file should look like this afterwards:
Switch back to the Calculation View Modeler and click the Expand Details Panel [1] icon in the top right corner.
Select Parameters [1] and you can see the three parameters [2] you added directly to the XML file.
Close the Details Panel.
On the Calculation View Editor click on the Join [1] icon and place the new node below the Aggregation [2] node.
Right-click on the Join_1 node and select Rename. Rename it to Join_TweetsSentiments to indicate that we join the Tweets with the Sentiments.
Select the Join_TweetsSentiments node and click on the Plus [1] sign.
In the Find Data Sources dialog search for Tweets [1] and select both views (TweetsSentiments and Tweets) [2]. Afterwards click on Finish.
!! CAUTION !!
Pay attention to the names shown in the results table. You will find the views from exercise 1 in the "Synonym" column.
As a result you will find both views in the Join_TweetsSentiments node. Click on the Expand Details Panel [1] icon next to the join node.
NOTE
If you didn't close the details panel in the steps before, it is already open on the right side of your screen.
In the Join Definition tab, drag & drop the TWEET_ID [1] property from the left table to the right. Change the cardinality [3] from m:n to 1:1. Keep the other settings as default, e.g. INNER as Join Type.
NOTE
You can find the cardinality setting in the "properties" [2] section. In case the properties is closed, you have to open it.
Switch to the Mapping section.
Drag & drop the columns listed below to the output columns.
- Tweets data soure: TWEET_ID, PLACE, CC_DYNAMICBASKET, CC_DYNAMICBASKET_DISPLAY
- TweetSentiments data source: NEUTRAL, WEAK_POSITIVE, WEAK_NEGATIVE, STRONG_POSITIVE, STRONG_NEGATIVE
NOTE
The order you perform the drag & drop of the nodes doesn't matter. You can e.g. start with the whole TweetSentiments node, delete the not required columns in the output columns and finally add the Tweet columns.
Switch to the section Calculated Columns [1]. Click on the Plus sign [2]. It will automatically add a new entry named CC_1 [3]. Click on it to open the details.
Change the Name [1] from CC_1 to CC_POSITIVE and the Data Type [2] to DOUBLE.
Open the Expression [3] section, change from SQL to Column Engine [4] and enter the following calculation statement "STRONG_POSITIVE" + ("WEAK_POSITIVE" * 0.5) [5]. Afterwards click on the Validate Syntax [6] button.
Once the validation was successful, you'll get a confirmation message. Click OK.
Repeat the steps and create two more calculated columns with data type DOUBLE: one named CC_NEGATIVE [1] with the expression "STRONG_NEGATIVE" + ("WEAK_NEGATIVE" * 0.5) and one named CC_NEUTRAL [2] with the expression "NEUTRAL" * 0.5. Afterwards close the Details Panel.
Select the Join_TweetsSentiments node and select the Connect tool [1]. Drag & drop the arrow (Connect tool) from the Join_TweetsSentiments node to the Aggregation node.
There's a connection between both nodes now [1]. Click on the Expand Details Panel [2] of the Aggregation node.
NOTE
If you didn't close the details panel in the steps before, the button will not show up and the panel is already open on the right side of your screen.
On the Mapping section add the columns TWEET_ID, PLACE, CC_DYNAMICBASKET, CC_DYNAMICBASKET_DISPLAY, CC_POSITIVE, CC_NEGATIVE and CC_NEUTRAL to the Output Columns.
Switch to section Parameters [1] and click on the Manage Parameter Mapping [2] icon.
In the Manage input parameter mappings section click on the Auto Map by Name icon [1]. As the names are the same it will automatically map the parameters.
Click on Semantics in the Calculation View Editor and open the Details panel.
Choose the section Columns and change the Label [1] columns for CC_POSITIVE, CC_NEGATIVE and CC_NEUTRAL. In addition set the column CC_DYNAMICBASKET_DISPLAY to Hidden [2] and map the Label Column for CC_DYNAMICBASKET to CC_DYNAMICBASKET_DISPLAY [3].
Right-click on the calculation view tweetSentiments.hdbcalculationview and choose Build > Build Selected Files.
Wait until the build process is finished and click Database Explorer [1].
Navigate to the entry Column Views [2] and select and right-click the corresponding calculation view with the name "ain365.exercise.models::tweetSentiments" [3]. Finally choose Open Data [4].
Set the Input Parameters for TIMEFRAME [1] to the value 0 and keep the other settings untouched. Afterwards click on the Open Content [2] icon.
Select CC_DYNAMICBASKET_DISPLAY as Label Axis [1] and add the three measures CC_POSITIVE, CC_NEGATIVE and CC_NEUTRAL to the Value Axis [2]. You should see a simple bar chart visualizing the sentiments by calendar week.
Congratulations! You have successfully completed the second exercise.
Continue with Exercise 3 and create the second calculation view.
This project is licensed under the SAP SAMPLE CODE LICENSE AGREEMENT except as noted otherwise in the LICENSE file.