{"id":2356,"date":"2020-09-08T09:52:43","date_gmt":"2020-09-08T08:52:43","guid":{"rendered":"https:\/\/ec-projects.co.uk\/intersystems-uki\/?p=2356"},"modified":"2020-09-08T09:52:43","modified_gmt":"2020-09-08T08:52:43","slug":"machine-learning-success-starts-with-healthy-data","status":"publish","type":"post","link":"https:\/\/ec-projects.co.uk\/intersystems-uki\/machine-learning-success-starts-with-healthy-data\/","title":{"rendered":"Machine Learning Success Starts with Healthy Data"},"content":{"rendered":"

This article was originally published on the InterSystems global website (link<\/a>), written by Scott Gnau, VP of Data Platforms.<\/em><\/p>\n

Machine learning (ML) and artificial intelligence (AI) are becoming core functions of successful organizations, with data playing an increasingly important role in making or breaking these applications.<\/p>\n

The required fuel for effective and precise AI and ML is healthy, normalized, and comprehensive data. In today\u2019s competitive business landscape, organizations can garner valuable insights from analytics to better inform the critical business decisions impacting their bottom line. Basing these decisions off of healthy data is often the difference between success and failure.<\/p>\n

Poor or unhealthy data has been\u00a0estimated to cost<\/a>\u00a0the US more than $3 trillion per year. Data science teams that don\u2019t have access to enough data lack the necessary resources to build successful ML architectures and models, which can lead to potentially flawed systems and less-than-desirable outcomes.<\/p>\n

The Foundation of Healthy Data: Organization<\/strong><\/p>\n

Healthy data is without duplicate records, missing information, formatting errors, incorrect information, or mismatched terminology. As the amount of data being generated and made available to organizations increases exponentially, collecting large amounts of data in an organized way can be a significant challenge. Failing to keep pace and properly collect this flood of data often results in missed insights and faulty outputs.<\/p>\n

Organizations looking to optimize healthy data must think beyond business objectives and take an integrated approach, identifying what data will be collected \u2013 including from where and how \u2013 and cleaned, and how its use maps back to the desired business goals.<\/p>\n

Healthy Data Benefits<\/strong><\/p>\n

Healthy data leads the way to better insights into all areas of the business. But even with the most cutting-edge models, a small piece of unhealthy data can significantly warp results. Healthy data has the ability to boost the effectiveness of an entire organization by providing:<\/p>\n