Northwestern Data Scientists Favor Plexa Over Hand Coding
Low-code paradigm is shaping the future of Data and Analytics, arming organizations with high productivity and low total cost of ownership. Low-Code in Data and Analytics is following in the footsteps of predecessor, Low-Code Application Development (LCAD). Gartner estimates that by 2024, Low-Code development will be responsible for more than 65% of enterprise application development activities. According to Forrester Research, the total spending in this category is forecasted to hit $21.2 billion by 2022, representing a compound annual growth rate of roughly 40% (Forbes). The race for disruptive innovation is on. Following the award of landmark US Patents, Kavi Global decided to pit their Low-Code, Technology-Agnostic Data and Analytics Platform Plexa against professionals.
Five data scientists from Northwestern University’s Masters in Analytics Program (MSiA) 2019 class took on the challenge during their Capstone project. Each data scientist was randomly assigned 2 Healthcare Data and Analytics use cases to solve using Plexa and hand coding. Scientists compared the two approaches in random sequence order across 6 Phases of the development lifecycle (Exploratory Data Analysis, Data Preparation, Model Development, Model Tuning, Model Validation and Model Accuracy), and ranked their satisfaction. Comparison criteria included User Experience, Capability, Functionality, and Productivity (time taken for development and documentation).
Plexa won in 3 of the 6 Phases (Data Preparation, Model Development and Model Tuning), tied in 1 (Model Accuracy), and hand coding won in 2 Phases (Exploratory Data Analysis & Model Validation). “This study made me a believer in the potential of Plexa and the commercial implications of empowering Citizen Data Scientists” revealed Dr. Diego Klabjan, Director of Master of Science in Analytics & Center of Deep Learning at Northwestern University. “Our team loved Plexa and provided recommendations to improve the Platform in the 2 Phases where hand coding won. We are planning to conduct another 9-month long exhaustive study with incoming Citizen Data Scientists to our Master’s program, starting in the fall quarter of 2020”, Dr. Klabjan added.
For this study Northwestern Data Scientists used a large, clinical dataset, developed by MIT Lab, named MIMIC II – Medical Information Mart for Intensive Care II. This dataset includes deidentified Healthcare data consisting of Demographics, Vital Signs, Laboratory Tests, Medications and more. The data includes 40,000+ unique patients over 60,000+ ICU stays at Beth Israel Deaconess Medical Center from 2001to 2012.
The 10 healthcare clinical use cases solved using Machine Learning and AI techniques during this study include predicting:Readmission, Discharge, Number of ICU Stay Days, Acute Kidney Injury, Chronic Heart Failure, Adult Sepsis, Days to Mortality, Mortality by Drug Adverse Events, Pulmonary Embolism, and Pneumonia.
Plexa is an integrated Development, Deployment and Operations Management environment for Data and Analytics. Being Low-Code, Plexa’s visual development paradigm offers compelling short-term productivity gains delivering speed, quality, and cost advantages. These capabilities are critical to efficiently manage the solution lifecycle, improve business-IT collaboration, and realize high-value business outcomes. “In addition to being low-code, Plexa’s patented design is technology agnostic, freeing our customers from vendor-lock-in and future-proofing Data and Analytics Solutions. These capabilities result in long-term risk mitigation including:technology obsolescence and skillset obsolesce, eliminating complex coding requirements and empowering Citizen Data Scientists” explains Srinivasan Chandrasekaran, Head of Software at Kavi Global.
Out-of-the box Data Engineering capabilities of Plexa include Data Ingestion, Transformation, Business Rules, Data Catalog, Data Quality, Scheduling, and Administration. Out-of-the-box Analytics capabilities of Plexa include:Data Prep, Feature Transformation, Feature Extraction, Sampling, Data Mining and Predictive Analytics, including dozens of popular algorithms in Machine Learning and AI. The Northwestern study was performed on Amazon AWS Cloud Platform. Plexa is also available for On-Prem deployment or on other major cloud platforms including Microsoft Azure and Google GCP.
“We are grateful to the Northwestern team for conducting the study and providing us with valuable insights and expert feedback for further improving Plexa. We value our partnership and the ongoing collaboration with Northwestern University”, says Vijitha Kaduwela, Founder and CEO of Kavi Global.