Brightspace Analytics Core

Editor's Notes

• #4: Things such as: How our Core Analytics provides a lot of what clients are looking for in terms of understanding: Course effectiveness (in-line tool stats) Being able to track Class and User Progress As well as some basic System level reporting And then there is how our Predictive Analytics (S3) provides the benefit of an early intervention system – by identifying students at risk and providing actionable insight to instructors embedded right within their regular workflows And then we have Insights which provides a more strategic view – allowing institutions to dig in deeper and get access to their data
• #6: Completion tracking Additional investment: Quiz event logs
• #7: PII - Deidentifiy information (before) - Has same PII as LE So now let’s take a look at our new BDP architecture: First big change – this is an event driven architecture. We have moved away from daily ETL to streaming events No more 24 hour data lag. Events are triggered in Brightspace, passed through an event connector (DEF), pulled into the BDP, processed and available within one hour. Not only providing near real time access to that data so that is available at the right time when the user needs it to act on it but also…. It enrichs the data… no more records – event driven A good example of how event driven data enrichs the data is… (talk about how LE only tracks last time user access a tool, accumulated amount of time in a tool and number of times access tool – but not full audit log as not good to track in a transaction production DB. And because originally used ETL’s which are just copy of records as they exist in transactional DB – original Insights also did not have ability to have a wider breath of data to analyse in regards to user audit logs. But with events that is all changing as each event is fired off and stored uniquely in the BDP which opens the door to better analysis of that data and better reporting opportunities (i.e. query date range to see all the times a user access a particular tool in that date range, etc). Second – Brightspace is not the only data source for populating the Brightspace Data Platform. BDP is Caliper ready and is able to consume caliper events from third party systems. Caliper is a standard created by the IMS Consortium (which D2L is a member of the board) that provides the interoperability between tool consumers like Brightspace and other tool providers to pass learning event information. LTI example – how Caliper fills the gap So now the BDP will store a holistic picture of the learning data relevant to a learners success. Third – the Brightspace Data Platform itself is built as an enterprise class multi-tenant cloud hosted infrastructure powered by a big data technology platform called Hadoop – which is open source framework for the distributed storage and processing of very large structured and unstructured data sets. Combined with other tools, Hadoop provides mechanism for storing and mining massive amounts of data extremely efficiently. So we have covered how data gets into the BDP, but how does data get out? How can you access all of that data? It is surfaced through a API service layer. The BDP is BI-tool agnostic and the data can be surfaced through an API service layer either through ODBC or REST-ful API. It is also really important to distinguish here that the BDP is NOT Insights. The BDP can be used in many ways and by many products – both within Brightspace and by third party BI tools owned by our clients. So let’s look at some examples of how BDP data can be used: One of the ways we are surfacing data captured within the BDP is through our Event Driven Reports within Brightspace Insights. The data behind the Event Driven Reports within Brightspace Insights – are pulled from the BDP via REST-ful Another way in which we are surfacing the rich data captured within the BDP is through embedded analytics within Brightspace via REST-ful API. Examples: new features to User Progress. But the point is this new architecture is opening up endless possibilities for what we can do to enhance both Brightspace itself as well as our Analytics Suite (Core Analytics, Insights, S3 and whatever else comes in the future…) 3rd Party BI-agnostic integrations – do your own via API or ODBC enabling the ability for you to do your own reporting as well as integrate Brightspace data with other data sources in your own enterprise reporting system – providing an enterprise view __________________________________________________________________________________________________________________________________________________________ BDP Technology for Additional Information (for Internal Use Only – do not share technology stack externally except to mention we are using Hadoop as our data store and that it is No SQL): Hadoop – open source framework for the distributed storage and processing of very large structured and unstructured data sets. It’s not the whole ecosystem as many think, and does require specific tools to interact with the data such as: Hbase – column-oriented database management system layered on top of the Hadoop DFS (HDFS) that enables analysis of the sparse data sets. It’s not relational, so doesn’t support SQL-like operations Phoenix – provides a relational database layer on top of Hbase – enabling SQL-like functionality for interacting with data Spark – open source in-memory clustered computing framework – loads frequently used data into memory to speed access Zookeeper – centralized configuration management service for distributed Hadoop clusters No SQL – method for interacting with data modeled in a way that’s not relational – columns being one example
• #8: BDP improvements in core