The journey so far... Our blog series began with: "Starting your Fusion Analytics Journey" which can be found here. In the second blog post, we discussed the various ways to move to delivering deeper and richer insights by bringing in non-Fusion data and combining with Fusion data which can be found here.
Next up…we will discuss the various ways in which we can further leverage the power of Fusion Analytics.
Using the self-service Machine Learning (ML) capability of Oracle Analytics Cloud (OAC), a Business Analyst can undertake ML against the data they have access to. This may be just Fusion data, or a combination of Fusion data and data from other sources.
We have seen Figure 11 in a previous blog, it shows a scenario where the Business Analyst has connected OAC to one or more live data sources. A Business Analyst with access to all this data can build, train, and deploy ML models within OAC.
To follow the HR example on attrition used in our previous blog, they could build and train a model to predict who is going to leave. This model can then be deployed in the front end to allow them to predict who is going to leave on an ongoing basis (note that models often need rebuilding and retraining to ensure accurate results).
You get full use of the capabilities of Fusion Analytics’ underlying Autonomous Data Warehouse (ADW) and as such, your Data Scientists have access to an extremely powerful library of ML algorithms. Figure 12 shows a Data Scientist taking advantage of the ML capability within ADW on top of the data that has been brought in from Fusion within the data pipeline along with the data coming in from other sources. Data Scientists can build comprehensive and robust models that they can then deploy into OAC dashboards such that the predictions generated can be consumed by the end user. A considerable number of blogs and information is available on this topic if you wish to explore the subject of Data Science and ML interest you.
We provide you with the choice of allowing the Business Analyst or IT and its Data Scientists access to tools to undertake powerful analysis. It maybe that the Business Analyst tests several theories and then asks IT to refine and provide a more robust framework to their work for delivery to the masses.
In our HR example, if your business requires ongoing prediction of who might leave, then you may wish IT to build a more comprehensive and governed attrition-model based on a larger number of attributes and larger volumes of data, which is then deployed into OAC and the results of which are then consumed by the business to better predict leavers.
In a real-life healthcare example, ML was used to understand attrition among nurses, looking at demographic factors, internal mobility, scheduling data, commute times, employee surveys. It was discovered that nurses traveling from the border towns of the state were leaving most often due to long commute times. By simply reallocating these nurses to more rural facilities closer to the state border, the hospital reduced turnover in this group by 70%.
The possibilities are endless, for example Oracle also provides AI Vision services, details here, as part of Oracle’s Cloud Infrastructure (OCI), upon which Fusion Analytics is built, it is an AI service for performing deep-learning–based image analysis at scale. With prebuilt models available out of the box, developers can easily build image recognition and text recognition into their applications without machine learning (ML) expertise. Therefore, it is possible to consider leveraging such advanced features within Oracle Analytics component of the Fusion Analytics framework.
A question often raised is “what role does Fusion Analytics play in the wider Data Strategy?” such as a Data Lakehouse (DL) Strategy.
It is important to remember that Oracle has applied its considerable knowledge - both technical and functional (of business) – and has invested a substantial number of man-hours into developing the pre-built solution Fusion Analytics. It has built a sophisticated data pipeline that enables the data to be curated and delivered into a ADW data model and exposed within OAC via a semantic layer built to accelerate the delivery of insights to a Business Analyst. In addition, as a SaaS product, Oracle manages Fusion Analytics, ensuring it continues to work as Fusion is enhanced, it is upgraded with new content and new technical capabilities. This not only reduces your development, implementation, and on-going management costs, it also de-risks every step, leaving you to worry about running your business, not the technology.
Bringing a Data Lakehouse (DL) strategy into the picture, with our first scenario, we see Fusion Analytics alongside a DL, Figure 13. Here, Fusion Analytics provides analytics for the Finance, Procurement and Supply Chain departments, whereas the DL is focused on providing data for analysis to other parts of your organisation – there is no need for the data to be combined for analysis.
Secondly, we sometimes see Fusion Analytics used to take advantage of the pre-built data pipelines, the curated and derived data in the ADW data model before moving it to their DL, Figure 14.
Thirdly, we often see a combination of the two, where the back-office departments have access to all out-of-the-box benefits of Fusion Analytics, whilst it also simultaneously feeds their DL for analysis by others, Figure 15.
Figure 16 simply shows customers using OAC within Fusion Analytics as the front end for the Data Lakehouse (sometimes this may be a separate instance of OAC).
In our ongoing HR example, the Business Analyst can marry the date in Fusion Analytics with data from the Data Lakehouse to undertake their analysis of why people are leaving, or IT and their Data Scientists maybe developing Machine Learning (ML) models within the Lakehouse and deploying those back into FAW’s OAC for consumption by the Business Analyst.
As mentioned previously in this blog series, Fusion Analytics is built upon Oracle’s Cloud Infrastructure (OCI) using two key services: ADW and OAC. Oracle has a complete set of services on OCI to allow you to deliver your Data Lakehouse with it. Doing so would allow you to leverage not only services such as ODI (ETL) tool, but Data Flow, Object Storage, Data Catalogue, Big Data Service, and the AI Services in Oracle’s cloud. Learn more about the Oracle Data Lakehouse here.
Figure 17 shows Fusion Analytics playing its part in an Oracle Data Lakehouse, with the Oracle Data Lakehouse taking advantage of the curated, derived data in ADW’s data model.
The combination of scenarios that we have covered so far and will cover in the next blog are not mutually exclusive and as you progress on you journey, you will see the ability to take advantage of all the capabilities of the Fusion Analytics platform, Figure 18 builds upon this idea.
At this point, it is worth highlighting that FAW is built upon OCI and some of its services, OAC & ADW for example, and thus FAW can be seen as simply part of the Oracle Data Lakehouse eco system, Figure 19.
In this blog, we have explored how Fusion Analytics can put Machine Learning (ML) into the hands of the Business Analyst within Fusion Analytics’ OAC, and how IT and the Data Scientists can play an important role with respect to ML and leverage the power of Fusion Analytics’ ADW. And we have also considered the value Fusion Analytics can play in the wider architecture of a Data Lakehouse. Read the final blog “Further Accelerating time to insight Analytics Platform” in this series here.
Duncan Fitter works in Oracle's analytics product strategy team helping customers across the globe achieve their goals through greater insights.