IBM Cloud Private for Data: Make Your Data Ready for AI & Cloud

Loren Murphy,
WW Analytics Technical Enablement Specialist
IBM

This week I traveled to Mexico City for a “teach-the-teacher” IBM Cloud Private for Data (ICP4D) training led by Dwaine Snow, Global Solutions Leader - Governance, Big Data, Analytics and Data Science. IBM technical sellers from across South America and Latin America gathered to learn how ICP4D can help customers be successful in their journey to Artificial Intelligence (AI).

What is IBM Cloud Private for Data?

ICP4D is an all-in-one Analytics and AI solution providing secured, governed access to your data and insight- across any cloud, public and/or private environment. It is built on Kubernetes and is underpinned by an open ecosystem. This enables you to build once and run anywhere. ICP4D helps you build and manage your Machine Learning and AI models, and it can run in your own data center or a cloud infrastructure of your choice (IBM SoftLayer, Amazon AWS, OpenStack, Red Hat OpenShift and coming soon Microsoft Azure).

ICP4D is unique because it was designed to create an end-to-end collaborative analytics and AI platform that targets four personas: Data Stewards, Data Scientists, Business Analysts, and App Developers.

Why is ICP4D Important? – The Digital Disruption

"75% of large enterprises will have digital transformation at the center of corporate strategy within two years"

Take a moment to think of companies who are disrupting their industry. At the heart of the disruption is data and the AI that is applied to that data to make the right decisions and recommendations. For example, Netflix transformed from a DVD distributor to now using all the data they have on movies, clients etc to provide personalized recommendations. Waze is able to understand traffic speeds, accidents, tolls and crowd sourced reporting events to determine the best route to get to where you want to go. And Uber Eats uses an algorithm to understand when to place an order with a restaurant, so it is picked up and delivered at the right temperature.

All companies have the opportunity to disrupt their markets with Data, AI, and the right platform but that digital transformation will require a data-driven culture.

Barriers to AI Transformations

"Data systems don’t “do AI” and AI technologies don’t “do data”

Regardless of the market or industry, there are often 4 common barriers to successful AI Transformations:

Data Ecosystem: Data resides in silos and is difficult to access. Unstructured and external data is difficult to consume. Governing the data and providing data lineage is extremely limited.

Analytics Tools: Discrete tools present barriers to productivity. Each department has its own preference and is using different tools which become difficult to manage and standardize.

Workflow:Workflows across the enterprise are not integrated or governed; thus, causing a lack of parity between development, testing and production which reduces productivity and increases time to market.

Culture: Teams often work in silos and are not collaborative across and within the team. Provisioning new services can be slow and stakeholders may lack trust in AI.

Why ICP4D – The Value

ICP4D addresses these common barriers to a successful AI transformation by providing a collaborative environment that focuses on the 4 main tasks in a data scientist or business analyst’s life cycle: Collect, Organize, Analyze, Embed.

Collect:

Collect relevant data and make it simple & accessible.

ICP4D enables you to connect to existing data sources, on-premise, off-premise, and/or upload your data.

Organize:

For effective AI, clean data is the price of admission. Decisions made from bad data cost the US economy roughly $3.1 trillion dollars each year and data-related challenges are hindering 96% of organizations from achieving AI.

With ICP4D, data stewards can auto-discover, catalog and publish meta-data. This enables data scientists and business analysts across the organization to “shop-4-data” in order to easily find the data needed to create AI models and dashboards. Data Stewards can also define and enforce governance policies and rules to ensure the integrity and quality of the data. Thus, improving the confidence of stakeholders and eliminating “garbage in, garbage out.”

Lastly, data engineers can build and execute advanced Extract, Transform, Load (ETL) jobs at scale to provide stakeholders the data needed in the correct format to build accurate models.

Analyze:

Analyze insights on demand.

With ICP4D, business analysts can easily build visualizations and dashboards to find insights and discover hidden patterns in the data which can then be communicated to the data science team for further analysis.

The Data Scientists can then create AI models using a method of their choice. The platform provides 4 ways to create models: Jupiter Notebooks, R-Studio, SPSS Modeler Flows -- a visual, drag-and-drop environment; and Watson Machine Learning-- an automated model builder. Once a model has been created, tested and deployed, its performance can be easily monitored. If performance degrades to below a certain threshold, the model can be re-trained. This is a differentiator for ICP4D because most AI tools only provide model deployment and not model management.