Data Science at a Glance

Introduction

The year 2020 has been both strenuous and transformative for society. From Air travel restrictions, to state-wide lockdowns, industries have adapted to survive new terrain. The unknown landscape has produced a 25% increase in in-home data usage from March 2019 to March 2020 (J.Clement, 2020). As a result of the substantial growth in data usage from customers, data science is more vital than ever. Enabling companies to provide informed decisions for to drive their services to best reach their customers.

Time in Sohosquared

While with Sohosquared, I have worked with various companies tackling different customer bases and, their various data needs. When advising our clients, we discuss what goals they need to achieve. From that discussion, we are then able to customize data requirements that focus on three clear goals:

  • Constructing Data Schemas for use of Data Science and Machine Learning tools of our products.
  • Reducing the collection of unnecessary data that would not assist in achieving the client’s goals. In essence, we reduce the bloat of data being collected.
  • Ensuring data security through all endpoints of the data flow.

Having worked with a scientific research and large multi-national financial company to a credit card company, we have been able to achieve the client’s success with their business needs at the forefront. Our clients are able to understand and own their data capture procedure, management, and analysis.

Trends of Data Science in 202

Looking forward, here are some trends that the industry is investing to increase customer interaction with their products.

  • Incorporation of common software practices such as continuous integration (CI) and continuous delivery (CD) as a standard operating procedure in building data products. This allows for products to react to dynamic and competitive markets in which the client needs to quickly adapt to remain competitive.
  • Continuous Intelligence or “Real-time” insight into your data. Given the data product life cycle follows the procedure of acquiring data, extracting information, gather insights through models, create decisions and enact on those decisions, the continuous intelligence allows companies to rank trends in terms of impact and make better decisions in their dynamic markets.
  • Attention to concerns of bias or fairness in Machine Learning (ML) or AI applications. Given that AI applications are becoming more ubiquitous in our daily lives, more pressure has been placed on companies by customers and government bodies to have companies to provide details on decisions are formulated by these ML products. Having worked with ML models, there are some that are considered black boxes in terms of how a human would describe their decision-making process. Libraries such as “LIME (Local Interpretable model-agnostic explanations) and “SHAP (Shapley Additive explanations)” serve as bridges to achieve the goal of having explainable models.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: