The Doppler Quarterly Summer 2017 - Page 11

A key component of the transition to becoming a data driven organization is the acceptance of a constant state of change, with a rigorous measurement compo- nent. These constant changes involve the automation of formerly manual tasks, requiring human approval, to a state where machines and algorithms make deci- sions and execute them on their own. This constant state of change is managed through common best practices (Figure 1) that assist an organization in ensuring high quality decisions are made. These best practices are driven by data acces- sible to a wide portion of the organization to greater facilitate collaboration. Maturity Level People Skills Process Methods Example Technologies Level 1 - Data Basic IT/Computer By memory Spreadsheets Access skills Level 2 - ETL, DBS By experience RDBMS Level Data Quality Documented & Enterprise Data Warehouse 3 - Reporting Statistics reproducible (Redshift, BigQuery, SQL Data Consolidation Development Level 4 - Alerting Advanced Warehouse) Automated Statistics Level 5 - Engaging NLP, Predictive, Big Data Platforms (EMR, Data- pric, HDInsights, Sprak) Learning & Evolving Modeling, Math Predictive Analytics Tools (R, Python, AWS ML, Google ML, Azure ML) Figure 2: Organizational Data Maturity Figure 2 illustrates the common maturity levels an organization will progress through as they become a data driven organization. An organization will not seamlessly move from one level to another, but rather mature each portion of the organization at different rates, depending on the skill sets and outside influence. Description of the data maturity levels: 1. Data Access – This is the first level of data maturity, and characterizes organizations that are early in their data journey. These organizations often store information for reference, but do not regularly use that infor- mation to drive decision making or work to integrate the data into third party systems for automation in use. 2. Consolidation – This stage in maturity characterizes organizations that have taken initial steps to integrate their separate datasets and create more formalized applications for the presentation and updating of the information. Decisions are still manual and human driven. SUMMER 2017 | THE DOPPLER | 9