Download the authoritative guide: Cloud Computing 2019: Using the Cloud for Competitive Advantage
Predictive analytics looks forward to attempt to divine unknown future events or actions based on data mining, statistics, modeling, deep learning and artificial intelligence, and machine learning. Predictive models are applied to business activities to better understand customers, with the goal of predicting buying patterns, potential risks, and likely opportunities.
Of all the forms of analytics, perhaps none is riskier than predictive analytics, because it is essentially fortune telling, though a highly sophisticated version. Business Intelligence, its predecessor in analytics, is a look backward. Who were our best customers? What were slow sales days? Yet in the era of cloud computing, this backward look is no longer sufficient – hence the market demand for predictive analytics tools.
Predictive analytics is reflected in today Big Data Trends, and its tools are essentially Big Data Technologies. The market demand for predictive analytics software corresponds with a closely related toolset, Big Data Analytics Tools.
Common uses for predictive analytics include but are not limited to:
- Optimizing marketing campaigns to determine customer responses to marketing campaigns or purchase patterns.
- Improving operations to better manage inventory and other resources, or to set prices for services based on things like seasonality.
- Fraud detection. Analytics can monitor activity and note or catch unusual or out of the ordinary customer activity, often in real-time.
- Reduce risk. Merchants, such as car dealers, use more than just a credit score now to determine whether to approve a loan. They also look at things like insurance claims and driving records to determine if the buyer is a risk.
Examples of Predictive Analytics
Each industry and sector puts predictive analytics to work in different ways. We break them down by industry and use case.
Probably the largest sector to use predictive analytics, retail is always looking to improve its sales position and forge better relations with customers. One of the most ubiquitous examples is Amazon’s recommendations. When you make a purchase, it puts up a list of other similar items that other buyers purchased.
Much of this is in the pre-sale area – with things like sales forecasting and market analysis, customer segmentation, revisions to business models, aligning IT to business units, managing inventory to account for seasonality, and finding best retail locations. But it also acts post-sale, acting to reduce returns, get the customer to come back and extend warranty sales.
One early attempt at this was Google Flu Trends (GFT). By monitoring millions of users’ health tracking behaviors online and comparing it to a historic baseline level of influenza activity for a corresponding region, Google hoped to predict flu patterns. But its numbers proved to be way overstated, owing to less than ideal information from users.
But there are other uses, such as predicting epidemics or public health issues based on the probability of a person suffering the same ailment again. Or predicting the chances of a person with known illness ends up in Intensive Care due to changes in environmental conditions. It can also predict when and why patients are readmitted and when a patient needs behavioral health care as well.
The most famous example is Bing Predicts, a prediction system by Microsoft’s Bing search engine. It has scored in the 80 percentile for singing contests like American Idol, the high 90s percentage in U.S. House and Senate races, and went 15 for 15 in the 2014 World Cup. It uses statistics and social media sentiment to make its assessments.
Another example is what’s known as “Moneyball,” based on a book about how the Oakland Athletics baseball team used analytics and evidence-based data to assemble a competitive team. It abandoned old predictors of success, such as runs batted in, for overlooked ones, like on-base. It took the Athletics to two consecutive playoffs.
Weather forecasting has improved by leaps and bounds thanks to predictive analytics models. Today’s five-day forecast is as accurate as a one-day forecast from the 1980s. Forecasts as long as nine to 10 days are now possible, and more important, 72-hour predictions of hurricane tracks are more accurate than 24-hour forecasts from 40 years ago.
The extreme polar vortex that dropped temperatures in Wisconsin and Minnesota to -50 degrees Fahrenheit was predicted several days out. All of this is done thanks to satellites monitoring the land and atmosphere. They feed that data into models that better represent our atmospheric and physical systems.
Despite some awful disasters in 2017, insurance firms lessened losses within risk tolerances, thanks to predictive analytics. It helped them set competitive prices in underwriting, analyze and estimate future losses, catch fraudulent claims, plan marketing campaigns, and provide better insights into risk selection.
Predictive modeling for financial services help optimize the overall business strategy, revenue generation, resource optimization, and generating sales. Automated financial services analytics can allow firms to run thousands of models simultaneously and deliver faster results than with traditional modeling.
It does this by analyzing strategic business investments, improve daily operations, increase productivity, and predicting changes to the current and future marketplace. The more common form of predictive analytics in financial services is the credit scoring system used to approve or deny loans, often within minutes.
Analytics in power plants can reduce unexpected equipment failures by predicting when a component might fail, thus helping reduce maintenance costs and improve power availability.
Utilities can also predict when customers might get a high bill and send out customer alerts to warn customers they are running up a large bill that month. Smart meters allowed utilities to warn customers of spikes at certain times of the day, helping them to know when to cut back on power use.
Social Media Analysis
Online social media is a fundamental shift of how information is being produced, particularly as relates to businesses. Tracking user comments on social media outlets enables companies to gain immediate feedback and the chance to respond quickly.
Nothing makes a local business jump like a bad review on Yelp, or makes a merchant respond like a bad review on Amazon. This means collecting and sorting through massive amounts of social media data and creating the right models to extract the useful data.
Alerting and Monitoring
This covers a wide range. Just in transportation, modern automobiles have more than 100 sensors and some are rapidly approaching 200 sensors. This gives a much more accurate report than the old generic Check Engine light.
Modern aircraft have close to 6,000 sensors that generating more than 2TB of data per day, which cannot be analyzed by human beings with any expedience. Machine learning to recognize normal behavior as well as signs leading up to failure can help predict a failure long before it happens.
Internet of Things
IDC estimates less than 1 percent of data generated today is being analyzed, and that flood will only increase as more IoT devices come online, such as smart cars.
Predictive analytics are needed to help sort what’s coming in to weed out useless data and find what you need to take intelligent actions. In one example, Cisco and Rockwell Automation helped a Japanese automation equipment maker reduce down time of its manufacturing robots to near zero by applying predictive analytics to operational data.