Category: Advanced Analytics and Data Science
Enbridge is a global energy infrastructure leader with natural gas and oil assets. Enbridge provides integrated services and first- and last-mile connectivity to key supply basins and demand markets.
Our project was designed to identify and understand the critical conditions that would allow us to accurately predict future maintenance events at wind farms. We needed recommendations to improve blade maintenance by increasing availability and reducing associated maintenance costs.
Our solution let us capture and store massive amounts of data from different sources and provide field inspectors with a mobile-enabled Web application to complete their blade inspections. Machine learning models were trained and verified to diagnose the defects and estimate a prognosis. Following a scientific process, a model based on a multiclass decision forest algorithm to classify blade defects was trained and deployed.
To predict the remaining useful life given the defects on the blade, a time-to-failure estimator was trained based on a Bayesian linear regression algorithm. The two models are responsible for understanding the severity and criticality of the defect. To make a maintenance recommendation, an automated decision model was deployed and validated. Overall, the solution let Enbridge capture, predict, and act on wind farm defects in a timely manner, resulting in 50 percent projected annual savings on preventive maintenance costs.
Category: Emerging Technologies and Methods
Ultra Mobile, Inc.
Solution Sponsor: Denodo Technologies
Ultra Mobile is a nationwide mobile virtual network operator (MVNO) in the U.S. that develops first-of-its-kind mobile phone services.
The company’s rapid growth was fueled by single sprint development that spawned fragile data sets. Although our information platform utilized modern technologies (e.g, Hadoop for data warehousing), the pace of application development exceeded our ability to integrate application data, limiting our ability to analyze the impact of change on our customers. Business was evolving much too quickly for us to build fact and dimension tables in the data warehouse. Querying the data for actionable insights often required multiple-page SQL queries that only a few data SMEs had the patience, skill, and knowledge to construct.
Because our data warehouse was maturing, often the SQL would reach into operational data sets and upset the balance between processing transactions and running strategic analytical queries. Our project's business objective was to enable data-driven decision making off the 14TB of data the business had generated in four years of operation. The team was given $76,000 and five months to execute.
In three months, the team’s implementation of the Denodo Platform enabled us to move from siloed, spreadsheet-based reports with little data lineage to a process-driven, governed, consistent, online view of data. This “freeing of the data” was transformational. We discovered quality-of-service issues and corrected them, stopping customer exodus. We could now audit TBs of transaction data, leading to profitable resolution of billing errors. We identified and corrected inefficient processes, resulting in our first-ever profitable quarter and continued positive trends.
The democratization of information through cultured data virtualization has enabled a first-year total investment of USD $128,000 to return a projected $2.6 million in revenue by the end of the year. The value of transforming data from a liability to an asset, without an increase in staffing, is priceless.
Category: Enterprise Data Warehouses
Solution Sponsor: Denodo Technologies
Asurion is a privately held Tennessee-based company that provides device protection and tech help services for smartphones, tablets, consumer electronics, appliances, and satellite receivers. It operates in 18 countries, serves 290 million consumers, and has 17,000 employees; 2016 revenue exceeded $5.8 billion.
As we expanded our operations worldwide, our customers expected a higher-quality customer experience. Simultaneously, we started to provide support services that extended beyond offering insurance and warranties. Asurion's new premium tech help service required strong predictive analytics, IoT capabilities, and big data architecture support to provide customers with the right experience.
Our on-premises legacy data architecture could not support the global expansion of our premium tech help service. We sought a next-generation data architecture that could enable our company to spin up additional infrastructure, services, and products in weeks, not months. We also faced strict restrictions on migrating data and had to remain compliant with stringent governmental regulations that required centralized companywide security management around a single point of control. To address this, we it embraced two technologies for a gen-2.0 data architecture: cloud and data virtualization.
We moved our data architecture to the cloud to be more agile, flexible, and cost-efficient as well as to support our international expansion. As part of the move, we needed to secure the cloud, implement proper access controls to ensure data integrity and data protection, and conform to contractual client-specific, regional, and departmental data rules. This cloud architecture allowed us to engage in better predictive analytics, data science, and advanced analytics while satisfying regulatory, privacy, and data security issues.
Category: Big Data
Solution Sponsor: Cloudera
Navistar, a leading manufacturer of commercial trucks, buses, defense vehicles, and engines, collects massive amounts of data from hundreds of thousands of vehicles and needed a highly scalable platform to conduct big data analytics. Previously, the company used a plethora of offerings to conduct analytics.
In 2013, Navistar released their remote diagnostics platform, OnCommand Connection, which leverages data feeds from telematics service providers and combines them with meteorological and geographical data, vehicle usage, traffic data, and historical warranty and parts inventory information to provide real-time vehicle performance data. To keep pace with the terabytes of data generated, Navistar chose Cloudera to build its big data solution. All the streaming data from sensors can now be directly ingested and combined with data from internal and external sources to drive insights and analytics, at considerably lower cost per terabyte.
With Cloudera, Navistar can perform ad hoc analytics and evaluate billions of rows of data in hours, not weeks. By tracking information from vehicles, Navistar can schedule preventive maintenance in advance of issues to reduce downtime and increase revenue.
Since partnering with Cloudera in 2014, Navistar has helped customers reduce maintenance costs and vehicle downtime by up to 40 percent and has implemented remote monitoring coverage for more than 300,000 vehicles on the road. Navistar also serves 3,600 buses transporting approximately 90,000 students every day, using IoT to access detailed information about school bus performance in real time, ensuring transportation conditions are safe and the buses are able to get kids to school on time.
Navistar’s use of Cloudera has allowed the company to apply data management and analytics practices that accelerate the company’s IoT and connected vehicle journey—driving critical business decisions that help Navistar further pave the way for the vehicle manufacturing industry.
Category: Visual Analytics and Data Discovery
Johnson & Johnson Consumer Companies, Inc.
Johnson & Johnson embraces research and science, bringing innovative ideas, products, and services to advance health and well-being. We employ over 130,000 people at more than 250 operating companies working with partners in healthcare.
Business Panorama is our next-gen insights and analytics platform that enables a path to growth across brand and customer teams worldwide by providing a one-stop shop for data and analytics solutions, tells compelling stories about our brands and business, and supports collaboration to drive actions. The solution enables a “connected systems” concept, putting business value and user experience first. Ongoing metrics around usability and value show that user base growth is accelerating. Users are gaining valuable insights, taking analytics from what to why. Increases in productivity result in value-added activities by users, further increasing the overall ROI of the solution.
Business Panorama was developed using a best-in-class mix of technologies, innovative concepts, and processes enabling a straightforward path to production, maximizing the leverage of opportunities with the organization, and expanding the user and organizational base. The platform seamlessly integrates the underlying technology layer into a portal that enables advanced visualization, data discovery, and insight generation at the right level based on user personalization.
Overall, Business Panorama enables Johnson & Johnson to harness massive amounts of data to further our mission of transforming health and well-being for people around the world.
Category: BI and Analytics on a Limited Budget
As the company behind the well-known U.K. job board, reed.co.uk, reedonline serves 15 million jobseekers and 50,000 U.K. recruitment agencies.
Our company was typical of many midsize and large enterprises with legacy reporting and stalled solo BI efforts. Our huge demand for insight was constrained by inflexible tooling, a creaking database (role-playing as a data warehouse), and access to a limited amount of data.
Data analysis was repetitive, manual, Excel-focused, and often limited due to the difficulty of acquiring data. Backlog items spanned weeks, sometimes months. We had to demonstrate the value of good data warehousing, BI, and analytics and transform our internal data offering.
We adopted a modern cloud architecture for both data warehousing and analytics complemented by on-premises ETL tools. Utilizing BI, analytics, and data warehousing technology, we now support a data-driven organization on a limited budget. We spent three months in vendor evaluation and procurement, followed by two months to roll out the first release to power users, then nine months to complete rollout to 300 users in our production environment.
We have 2 TB of data pulled from five core systems; we support two data warehouses and use five technologies. We have over 100 business-focused analytical models reusing common dimensions, about 30 centrally managed reporting solutions, and 100 different ETL packages. We enable data analysis across hundreds of metrics and thousands of dimensions. Our biggest fact table contains 600 million rows, giving us 10 years of insight (compared to 30 days before the project).
Only three legacy reports remain (rewritten to use our new data warehouse). Our analysts produce product- and customer-focused dashboards and reports; all power users can save their own reports, which are validated against our centralized data model to identify any breaking changes. Fifty percent of users utilize self-service though dashboards, 45 percent are power users with analytical and content creation capability, and five percent of users (outside the BI team) can add new dimensions, measures, and KPIs to our enterprise data models. Working in an agile fashion and as one virtual team, we can respond to new demands for insight in minutes or hours. We crowdsource requirements and empower analysts to evolve our models, ensuring that we remain responsive to diverse and dynamic business needs.