End-to-end analytics in the cloud
Data and analytics drive actionable insights. The industry is seeing a shift in the data and analytics space from point, discrete products to unified, cloud-based platforms. IBM has a comprehensive portfolio of offerings that deliver cloud-based, end-to-end analytics capabilities. These capabilities consist of cloud services for ingestion, persistence and store, analytics and visualization.
Cloud services can ingest a variety of data types—structured, semistructured and unstructured—at scale from different sources—enterprise, Internet of Things devices, third-party feeds and so on. Different data types can be persisted and stored in the cloud in a fit-for-purpose manner. A polyglot persistence of best-fit stores makes up a logical data lake in the cloud. We can apply a variety of analytics against this data, ranging from basic exploratory analytics to advanced, predictive and cognitive analytics. Different personas in the organization such as business analysts and data scientists can use role-appropriate tools to interact with, visualize and derive insights from the data.
Applying personas and analytics in a retail scenario
To demonstrate this end-to-end capability, take a look at this built-out scenario:
In the scenario, a retail business runs stores across the country. These stores carry different kinds of merchandise including golf, camping and outdoor protection equipment. Beacons in each store track customer presence. A business analyst looks at sales data and observes a decline in sales of outdoor protection products. But what is causing this decline? Could it be a change in customer interest? Maybe weather patterns? Perhaps an unpopular brand? Or could the decline be attributable to inefficient promotions?
A data scientist can perform advanced analytics by looking at sales data, presence data and weather data in context. Further, the data scientist can build models that predict demand based on customer attributes and conditions.
The scenario shows the use of IBM Cloud Services to ingest three different types of data. This process includes relational enterprise data using DataWorks, Internet of Things data from beacons in the store through the Internet of Things Foundation service, and weather data using the Weather Insights service.
The different data types are persisted in best-fit stores that constitute a data lake in the cloud. Relational warehouse data is best fit in dashDB, Internet of Things data is best fit in a NoSQL store such as Cloudant, and bulk weather data is stored in an object store. IBM BigInsights or Apache Spark as a service provides Spark analytics in the cloud.
Different personas interact with the data in different ways. A citizen analyst uses IBM Watson Analytics to get a guided, exploratory experience that yields insights. This analyst discovers not just the decline in sales for outdoor protection products, but also factors such as poor promotions that contributed to that decline.
A data scientist uses notebooks to do more advanced, interactive analytics. Spark analytics against all of the different data sources, in context, enables the data scientist to derive deep insights. The data scientist finds that weather patterns or aisle location did not play a part in the decline and confirms that a more targeted promotion has the potential for improving sales.
In addition, predictive analytics allows for the creation of models that predict demand. The end result is that the business is able to derive actionable insights that help correct the decline and optimize cross-sell and up-sell opportunities.
End-to-end capabilities reveal insights
Hands-on implementation of the scenario brings to light a number of advanced analytics strengths:
- IBM offers a cloud-based, end-to-end portfolio of analytics capabilities. This comprehensive portfolio spans data ingestion, fit-for-purpose data stores, analytics and visualization.
- It supports different personas and tooling that is appropriate for skills and roles in the organization.
- It focuses on Spark, its foundational technology. Spark connectors enable seamless analytics across a variety of relational and NoSQL databases and object stores in the cloud.
- Its cloud model helps make consuming analytics services extremely easy. The time taken to set up and implement the solution is dramatically lower than traditional approaches.
Learn more about this analytics portfolio. Register for a webcast presenting end-to-end analytics in the cloud with IBM.