Data Analytics Assessment | Center of Excellence Design
Standing up a successful data science team is a complex and error-prone undertaking. Data science differs substantially from ordinary IT or business intelligence work. The most effective data science practitioners are trained applied scientists – scientists with the training and intuition to reach practical business outcomes using scientific methods. Analytics are no longer a hope and a wish; all businesses need to adapt to the advantages data-driven decision making brings, or they risk being left behind. As seen in the figure below, advanced analytics are more challenging, but they also bring greater value.
A prominent energy generation and transmission firm was seeking to understand how data science and advanced analytics could help their business, as well as the level of investment and organizational change that might be necessary to take advantage of those opportunities.
The public utility wanted to focus on utilizing internal data for improved business decision making, optimizing their data analytics Center of Excellence (CoE) team structure, matching analytics technology with organizational fit, and convincing business stakeholders of the value and possibilities of advanced analytics. Mosaic Software (Mosaic) was contacted to perform this enterprise-wide analytics assessment. Mosaic brings over a decade of advanced analytics consulting experience and has helped many businesses ascend the analytics maturity curve.
What we did
Mosaic recommended the energy firm implement a repeatable data analytics process. A workshop was designed and delivered to focus on best practices around analytics. This session involved interviewing the different business unit leaders, IT and analytics leaders as well as an overview of what data science is, how it is optimally applied, and a high-level review of technologies/processes best suited to meet those objectives. Mosaic proposed implementing the Cross Industry Standard Process for Data Mining (CRISP-DM, as seen in Figure 2). This process breaks the analytics solution development process into six steps; business understanding, data understanding, data preparation, modeling, evaluation, and deployment. CRISP-DM emphasizes the iterative nature of analytics solution development with all development rooted in a deep understanding of the business requirements and objectives.
Armed with knowledge and business insights from the preceding phases, Mosaic established a set of guiding principles and built an optimal roadmap for people, process technology, and projects. The report delivered from this section mapped known advanced analytics skillsets and capabilities throughout the organization, cataloged current analytics toolsets in use, identified gaps with respect to realizing value from data science, highlighted recommendations and a roadmap for addressing gaps, and analyzed current feasibility of analytics use cases.
Following initial stakeholder interviews, Mosaic assisted the client in assembling a core team to serve as the conduit for continuing the discovery process, interacting with stakeholders, and collaboratively developing a data analytics strategy supported by stakeholders. This involved designing and delivering recommendations for a hybrid data management organization to include a CoE that would function in collaboration with the lines of business (LoB).
Mosaic developed and delivered a technology and skills matrix to satisfy each phase of the analytics infrastructure lifecycle.
Mosaic also helped outlined business processes for a high performing data warehouse that facilitates analytics.
The organization had begun to investigate deploying a data lake or a data warehouse. Mosaic was able to assist by taking them through the process of figuring out what the objectives were and ensuring results were on point. The data lake/data warehouse process required a holistic approach for all stakeholders, and this process necessitated collaboration between IT and business units, while requiring governance to ensure success. Timing and scheduling of operational workflows need to be designed with the end user in mind. The process must anticipate failure and include recovery tasks as appropriate along the entire chain. Processes must anticipate and accommodate change, especially from the data sources.
The Mosaic team was able to provide an objective third-party view of the technology market since neither firm is tied to any specific platform. Mosaic champions open source technologies such as R and Python for firms just starting out in their analytical journey. R and Python are also great tools for mature organizations and can easily be utilized with almost any data platform.
Upon completion of the rigorous analytics assessment, the utility had a comprehensive plan to leverage data analytics to inform decision making across the business. They now have a clear path forward on how to build a CoE, staff it with the optimal number and type of resources, and provide those resources with cutting-edge technology. Mosaic was able to draw upon its deep expertise, maintain an objective view of the utility’s enterprise landscape, and deliver a powerful, return-on-investment focused engagement.