Getting started with Cloud for Analytics: Part 1 – Creating an Analytic Model
Having had chance to get some hands on with Cloud for Analytics I want to create a small blog series to talk about the experience and showcase the steps and learning I have gone through.
Part 1 – creating a model for analytics
Part 2 – loading my analytics model with data
Part 3 – building a Story (dashboard) on top of the model
OK, so on with part 1 – building my model, which is the foundation of exploring, visualising and planning. After getting access to the system the first thing I wanted to do was create a dashboard (a Story in Cloud Analytics) based off an Excel file of data.
First I needed to create my model. There’s various ways you can create your model in Cloud for Analytics:
- From the Modeller area you can create a model manually (creating new or existing dimension) – this offers the most flexibility when creating your model. You need to load data into the model after you have created it.
- From the Modeller area you can create a model by importing it (from an Excel file, BW, BPC, HANA, HCP) – this will also load the data too. A good way to get started.
- From the Connect to Data wizard option on the homepage (allowing you to connect to an Excel or Google Drive). Again this will load the data as well as creating the mode.
- By creating a new Story and connecting to an Excel file or Google Drive at that point. Again this will load the data as well as creating the mode.
After trying option 2, 3 & 4 to begin with to get myself familiar with the modelling concepts, allowing with viewing the creating a planning model and creating an analytic model on the Cloud for Analytics YouTube channel I decided to go for option 1 and create a model myself. Option 1 allows the greatest level of control for your model, including the ability to choose the default currency (which in option 2 – 4 was selected as USD by default without the ability to change it).
To start with, from the Modeler section, press the create new model button.
First you need to specify a technical name and description and specify whether the model is enabled for planning (selected by default). As my mode was for analytics only I deselected this option.
Next I needed to specify the lowest time granularity, which in my case was Day, and the start and end date ranges (I had data from 2012 onwards).
I was then able to change the Preferences of the model and change the default currency to GBP. At this stage I left the other options as they were.
Back in the main model area, next up was creating the Accounts dimension and I need to create new one rather than reusing an existing one. The account dimension is essentially the table of measures/key figures that you have in your model – I guess you can think of it as the fact table. Every model must have an Account dimension (and only one).
After specifying a name and description for the dimension I was then able to manually key (or paste in) the ID, descriptions of the measures along with the unit types, aggregation behaviours and decimal places. My model didn’t require it but I can build a measure hierarchy if requires (referencing the row ID for the parent measure) and add a formula syntax if necessary. I made the unit for the Order Quantity QTY rather than Amount.
Next was then to create my other dimensions (i.e. my characteristics) by clicking on the plus button. Again I created a new dimension rather than reusing an existing one – the fact that you can re-use dimensions (and the values within them) is useful.
The first dimension I created was for Customer Segment. I choose the Generic dimension type before specifying a technical name and description. The dimension types are:
|Organisation||The Organisation dimension type is optional in a model but offers an organisational analysis of the account data, based, for example, on operating units, geographic entities, or cost centres. A model can have only one organization dimension|
|Generic||The Generic type is a free-format dimension to which data can be added or imported as required. This could be based, for example on products, channels, or sales representatives. A model may have any number of generic dimensions|
|Pool||The Pool dimension type is specifically for use with Allocation for planning models|
I was then able to specific the values of Customer Segments (IDs and Description). My data has the descriptions of segments so I didn’t need to map and ID and text. I pasted the values in. I could have added a hierarchy column if required.
Then I needed to create a new dimension of Organisation type for the City. I added in two additional columns to allow me to specify a longitude and latitude to allow me to use a Geo Map in the dashboard.
I then went through the process of creating a number of other dimensions until I was ready to finalise the model.
Hitting the save button saved my model before making it available on the main Modeler screen, ready for me to load data into.
There were a number of other options for building my model but this was enough for what I needed to get started.
Thanks for reading this blog. Next up will be Part 2 where I’ll discuss how I then loaded data into this model.
thanks for this blog, which is really useful to understand the key concept of model within C4A.
Thanks for the feedback Loic.
This sounds great! I have a question when you say "the fact that you can re-use dimensions". Can we reuse the BPC dimensions across different models in SAP Analytics for Cloud, because when I create a model using existing Datasource then in the mapping screen system creates dimensions COSTCENTRE_150820494940 some random number and then maps to the BPC dimension? Whereas in BPC I would be using the same Costcentre dimension in different models.