In the last five months, I worked closely with the business stakeholders in helping them understand the sort of insights they might want to extract from the raw data. This helped me to understand the importance of data sitting quietly. So for the first time institutions start making decisions based on data and rigor, instead of gut and hunch. I believe a prudish tool to get the basic insights from the raw data will radically shift the way we associate with our potential customer.
Without data, anyone who does anything is free to claim success - Angus Deaton
My companion in making this rigorous data mashing is Salesforce analytic cloud build on wave platform. I think Wave platform is treated in a very different way than any other product released by Salesforce in the last couple of years. There is little importance or communication given about it. As its very difficult to have your hands on wave, even the community is not very active in pushing this thing up.
Just to give a quick introduction of the tool, the Analytic cloud is the flagship product of Wave and categorized as 'Data Discovery Tool', It is a mobile-first solution that allows data to be sourced from multiple platforms, manipulated in real-time, and shared with other users using the wave platform. It used faceted data model where each column in the record becomes an attribute of the record and Indexes or build for every attribute, allowing rapid retrieval.
Building blocks of Wave
Datasets are created from CSV or any other source
Lens is temporary snapshot for the data with a defined measure and dimension
Dashboard contains system metadata and the dashboard state with tag name
, which contains sub-sections including Steps, Widgets, and Layouts
Apps allows you to bundle the dashboards, lenses, and datasets
My experience with these building blocks can be summarized in three different levels
You start with loading csv or pull Salesforce data as a data set, which is very neat and quick
Then you will have to deal with complicated scenarios where you have to transform and load some data sized more than 500Mb, which forces to switch from UI to API. This will let you sum up your ETL skills to push the data into Salesforce wave to create a dashboard and then you will modify Json files to augment multiple datasets into a single dashboard
Then swiftly you will fall into a category where you have to extract snapshots based on calculated insights defined by the business. For such a transformation you have to deal with Velocity and Volume. Even with out Variety of data you will switch to 'BigData' for processing information before pushing into Wave platform
Big data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it...
During this phase, I am totally surprised by the way how the complexity upsurges when you switch from UI to API. So I will share the tools I have used in loading the data and doing transformations in Salesforce Analytic cloud
Wave Labs App - https://wave-labs.herokuapp.com/
Dataset Utils - https://github.com/forcedotcom/Analytics-Cloud-Dataset-Utils
Limitations of Wave
Wave can only do basic aggregations on numeric data. It has no capability to perform text processing and it lacks visualizations for text analysis, like tag clouds. And Wave has no means of handling geospatial data in its visual controls or widgets. Advanced Math and statistical functions are also not available in Wave.