What Is Predictive Algorithmic Forecasting and Why Should You Care?

AI, machine learning, predictive analytics and algorithmic forecasting are constantly discussed in the mainstream media, but how do they lead to business success?

By Dipti Parmar

By Dipti Parmar October 17 2019

Many people start their days asking Alexa about the weather. They shoot out quick replies to emails with Gmail’s Smart Reply function. They drive to work using map apps that reroute each journey to best avoid traffic (and traffic cops). Once back home, they vegetate in front of the television as Netflix takes over control of entertainment needs.

Without even realizing it, the world is increasingly powered by AI. When asked if the world is better or worse since the widespread adoption of AI, 61% of respondents agreed in the affirmative, according to ARM’s Global Artificial Intelligence Survey.

Hyperconverged Infrastructure Adoption Chart

Source: ARM

Side note: Avoiding traffic emerged as a top desired application of AI among the nearly 4000 respondents in this global survey. Something for AI developers and AI-based startups to consider!

While consumer level adoption of AI and machine learning has been quite pervasive, research shows that many businesses are much slower to embrace this new technology.

Source: Gartner

One of the most significant areas in which companies and organizations can start leveraging AI is decision-making, because it can process a lot more data and information than humans and identify patterns and trends in consumer behavior.

Enter predictive algorithmic forecasting.

What is Predictive Algorithmic Forecasting?

Any form of “forecasting” essentially paints a scenario about the future based on current and historical data.

Predictive algorithmic forecasting refers to a method of AI-based estimation, where statistical algorithms fed with historical data make predictions on what is likely to happen in the future. As more data flows into the algorithmic model, the model automatically “learns” more about the scenario and its predictions get increasingly more accurate with time.

Unlike traditional statistical forecasting, which gives one set of results and rests on them, predictive algorithmic forecasting is not static. It becomes more precise and course corrects automatically based on the data flowing into its model.

However, AI isn’t the only aspect that matters to a good predictive algorithmic model. The most successful predictive algorithms are those that have a strong human intelligence component to them. It takes an astute data scientist to translate the findings from a predictive algorithmic model, turn them into recommendations, and apply them to a business scenario.

Source: McKinsey

This combination of human and artificial intelligence is what delivers the 1-2 punch setting predictive algorithmic forecasting head and shoulders above its traditional counterparts.

Applications in the Real World

Businesses can use AI-based technology to create value in two ways: scaling up or saving costs. Nutanix Xi Beam is a case in point. Beam’s AI-driven algorithms track customers’ cloud usage patterns, identify under-utilized resources, and proactively offer recommendations for cost savings.

Source: McKinsey

Amazon’s personalized product recommendations laid the foundations for its massive, loyal user base, thanks to its collaborative and content-based filtering features. Other examples of predictive algorithms at work exist in almost every industry, including finance, healthcare, transportation and consumer goods.

These kinds of predictive algorithms do impact businesses’ bottom lines just as they touch lives and make the world a safer, better place.

Healthcare and Suicide Prevention

In 2017, researchers at the Vanderbilt University Medical Center used algorithmic forecasting powered by machine learning to create a suicide prediction model that could be applied to incoming patients at all hospitals.

They used hospital admissions data for over 5000 patients, including age, sex, zip code and diagnostic history to build a model that could predict the likelihood of a patient being suicidal. The model predicted the possibility of a suicide over the next week, in the case of new incoming patients with an 84% accuracy.

What’s more, it continued to be able to predict the probability of a patient’s likelihood to commit suicide up to two years down the line to 80% accuracy, thus helping hospital staff and doctors from reaching out and offering timely support to prevent such a drastic step.

Social Services and Improving the Foster Care System

Machine learning algorithms were harnessed by the social entrepreneurship company DataKind in central Florida, to help foster care workers in the county improve their productivity and reduce caseworker turnover.

They built a tool that uses historical data to predict how many hours a new case will take. This helps in assigning the right number of cases to each case worker to optimize case load and prevent worker burnout.

It also helps case workers plan their activities for each week using past historical case data, maps and routing data from Bing Maps. Case workers can enter their case details into the tool and get an algorithmically-determined plan for their week, including meeting times, routes and visits without having to waste time planning, rerouting or canceled visits.

“I have seen route optimization make a 30-mile difference for a day’s worth of activities,” said Matt Baker, Director of Business Analytics and Automation at Community Based Care of Central Florida (CBCCF). “That can save hours a day and $2,025 a day in travel costs alone for 150 case managers.”

Insurance and Claims Processing Efficiency

Zurich Insurance uses predictive modeling and AI to automate a large part of its cumbersome injury claims filing and assessment process. The medical report evaluation process is fully automated; routine tasks like re-entering data into the system and updating databases are done by software, freeing up time for employees to focus on more value-added activities like negotiating settlement amounts with the claimants.

This combination of automation and AI tools has helped Zurich Insurance improve productivity without any impact on the quality of decision-making. Time taken to assess a medical report went from one hour to just a few seconds, saving nearly 40,000 hours of manual work and about $5 million in overall cost savings.

What’s Next?

Though we see the significant ways in which AI and algorithmic forecasting have changed the way we live and do business, somewhere there still exists a reluctance to consider AI as nothing more than a shiny new toy, whose real revenue impact is still nebulous.

However, this mindset flies in the face of real facts as revealed by this Deloitte study. A staggering 83% of early adopters of AI have already achieved moderate (53%) or substantial (30%) economic benefits from their AI investments.

These returns are only set to leapfrog with time. Accenture’s data shows that AI and algorithmic forecasting will boost profitability by 38% and generate additional revenues to the tune of $14 trillion by 2035.

The real question now is, how many SMBs and enterprises will grab sizeable chunks of that additional revenue by building predictive models and applying them in their everyday operations?

Dipti Parmar is a contributing writer. She has written for CIO.com, Entrepreneur, CMO.com and Inc. Magazine. Follow her on Twitter @dipTparmar.

© 2019 Nutanix, Inc. All rights reserved. For additional legal information, please go here.