As companies incorporate data science and machine learning to help them push into the new AI frontier, some are finding it difficult to get data science operationalized. We know organizations have data, and lots of it, but not every organization achieves the same level of agility from its analytics and data science initiatives, meaning one organization may see the impact of data science more quickly than another.
To better understand the issue around agility, we can look at the job of a data scientist. Data scientists are asked to find an explanation to a business question or solution to a business problem using various data sets (internal or external). For example, a question a data scientist may be asked from the Chief Marketing Officer (CMO) can be “How can we increase customer loyalty?”, a question that can now be answered using data and analytics. However, before an organization can get to the answer, its data scientists need to spend a considerable amount of time preparing the data – one of the main hurdles to achieving agility with analytics.
Currently, many organizations’ data scientists spend 80% of their time in the data preparation process, which includes collecting, cleaning, and transforming the data. The outputs from these steps are a crucial part of the data science process and lead to accurate and valuable models. However, even in the simplest situation too much time on these steps lead to a costly project, and in the most extreme cases by the time the data is in a usable format, it’s outdated, it doesn’t reflect the current customer landscape, and the opportunity to generate value has been missed. Therefore, one of the major hurdles an organization faces is its ability to be agile with analytics and data science. If that same CMO wants a quick answer to make timely business decisions that may increase customer loyalty, then the organization needs to create the internal capability to answer such questions quickly.
Adding Agility to Data Science and Analytics
One of the ways we, West Monroe Partners, have approached this challenge is to develop Rapid Analytics Platform (RAP), which is designed with the data scientist in mind. RAP helps eliminate the complicated data preparation tasks and processes, and it has the functionality to quickly and seamlessly add new datasets. For example, if an organization is licensing any form of 3rd party data, it can use RAP to quickly collect and ingest the data and as result will not have to worry about writing off the first few weeks of licensing costs as it tries to integrate the new data source. By using RAP as a data science enabling data platform, data scientists can quickly clean data, deal with missing values, apply business rules, merge and aggregate data, and easily create new datasets that hold the data they need without having to write code – simplifying the data preparation process.
For organizations that may not have a well-developed data science team, it may be challenging to ask analysts with some data science modeling skills to also take on the data wrangling responsibilities of data scientists if they don’t have strong technical knowledge to write ETL (Extract, Transform, and Load) code that transforms data into a usable format. We realize this is a challenge across many organizations, and we wanted to address that with RAP.
RAP and the Modern BI Architecture
To understand where RAP fits into the Modern BI Architecture shown in Figure 1, think of RAP as an accelerator that retrieves data hiding in every corner of your organization, retrains it to speak a unified language, and maintains a clean data engine even as volume increases. RAP is the modern, adaptive platform that enables the iterative, agile implementation of data integration and business analytics. RAP can help simplify and automate many data collection, data cleaning, and data transformation tasks that must occur in the analytics and data science development lifecycle. Then, through simplification and automation, analysts and data scientists can focus more of their time on model selection and improving accuracy, rather than writing cumbersome queries to extract and manipulate data.
Figure 1.0 – Modern BI Architecture: Forrester, 2017
Overall, Rapid Analytics Platform can cut the work of the data scientist by up to 80% by streamlining the data preparation process, giving organizations of any analytical maturity more agility to answer the pressing questions that only data can help answer.
Contact us and let’s discuss your analytics and data science challenges – and check back regularly for more updates around our business analytics advisory services.