Issue link: https://read.uberflip.com/i/1471966
BUILDING A MODERN STRATEGY FOR ANALYTICS AND MACHINE LEARNING SUCCESS 4 Inconsistent data governance. Some organiza- tions have pieced together governance policies over time, resulting in stove-piped access control policies that can lead to stale data, security vulnerabilities, and regulatory noncompliance. Other organizations have overcompensated with restrictive governance policies that impede productivity. "A common mistake we've seen is that organizations end up creating overly rigid governance scenarios, where the central teams become a bottleneck to those trying to work with data to improve the business," says Pathak. "The key to good governance is figuring out how to define access, then getting out of the way. By that I mean creating exception processes, rather than taking an approach that anytime you need data, you have to ask someone in a central organization for it." What's the right mix of analytics? There are many kinds of data analytics, and your business will likely want to apply a mix of them to achieve various outcomes. Here are the basic analytics types and their most typical use cases. Real-time analytics turns data into insights as it's being collected. This type of analytics is used for highly time-sensitive applications such as online trading or vehicle control systems. Real-time analytics can predict when equipment is about to fail, help a self-driving vehicle avoid an accident, and detect credit card fraud before a transaction is complete. There are two main types of real-time analytics: On-demand analytics waits for users or applications to send a query before delivering a result, while streaming analytics continuously delivers alerts or results. Log analytics, also called operational analytics, is the assessment of event data that might be captured from a computer, network, application, operating system, or another IT component. An organization can use log analytics to uncover patterns in user behaviors, identify trouble spots, audit security activities, manage regulatory compliance, and plan for capacity or other IT infrastructure changes. Big data analytics involves running ad- vanced analytics against very large, diverse data sets that might include structured, semi-struc- tured, and unstructured data from different sources and in different sizes. The data might originate from sensors, computing or communi- cations devices, video/audio, networks, log files, transactional applications, web content, and social media. Data warehousing analytics performs queries against large amounts of historical data gathered from many sources such as applica- tion log files and transaction applications. This type of analytics lets users run queries based on subject and assesses changes over time. Machine learning analyzes and interprets patterns in the data to enable learning and deci- sion-making without human interaction. It can help create entirely new revenue opportunities, enable better and faster decisions, and improve operational efficiencies.