FUNDAMENTALS OF PRESCRIPTIVE ANALYTICS

APPLICATIONS OF PRESCRIPTIVE ANALYTICS IN VARIOUS DISCIPLINES

Some of the areas where Prescriptive Analytics is currently providing

optimization benefits are:

1. Marketing. It is in this area that saw the earliest applications of PA.

Predictive analytics uses data models, statistics, and machine learning to
predict future events. In marketing, this can be used to make better
decisions regarding media planning
and buying. Using this tool, marketers
can gain a better understanding of
which campaigns are working and
what sorts of advertising will lead to
an increase in sales in future.

There are several ways that marketers can incorporate predictive

marketing analytics into campaigns to improve effectiveness and enhance
marketing:

1. Understand Consumer Behavior

The insights give marketers an understanding of consumer interests
based on past interactions. Marketers can segment audiences based
on known interests and demographic information. This better equips
marketers to serve individuals targeted messaging at the right time
on the right device. Ultimately this helps break through constant ad
noise — improving consumer experience and brand loyalty.

2. Optimize Resources and Spend

Marketers can determine where to focus advertisement spending
based on the value the customer presents. Predictive data identifies
the advertising channels and times that warrant increased marketing
spend and resources as well as where to reduce investments. For
example, data for an ice cream brand would likely indicate that
marketing efforts in northern temperate areas will yield better results
in the summertime, allowing them to save the bulk of the ad spend for
summer months.

3. Qualify and Prioritize Leads

Predictive analytics also allows marketers to qualify and prioritize
 leads. Insight into consumer behavior allows marketers to identify
ideal audience segments that are closer to conversion. These
analytics demonstrate how likely a consumer is to act, allowing
marketing teams to devote more attention to those consumers and
minimize wasted ad dollars on consumers that will not respond to
marketing.
4. Retain Customers
Finally, predictive analytics assist in customer retention efforts as
marketers are able to better understand consumers’ needs. This helps
with product and service offerings that complement consumer
purchase history and interests and also sheds light on cross-sell and
upsell opportunities that are not worth exploring.

Data insights in Marketing should also provide answers and action

options to questions such as:

- What customers want?

- How much cost did we acquire them?
- How can I acquire more customers like them?
- Which customers are the most loyal?
- How long will they stay with us?

Businesses reduce cost of running the business by interacting with their

customers and improve profit, so prescriptive analytics options should
provide insight on what interactions make their customers loyal.

2. Health Care. This is an emerging area for PA. The healthcare industry
is under pressure to deliver better outcomes due to increasing consumer
demand. This pressure is coming
from many directions, including a
shortage of primary care physicians,
the pervasiveness of chronic illnesses,
and the increased cost of healthcare.

In response to this pressure, many

healthcare providers and insurers are moving from a volume-based
business model to a value-based business model. This shift improves
outcomes by making healthcare more efficient and effective. However, it
also creates new challenges for healthcare professionals, who must now
be more productive and efficient in their work. The extreme pressure to do
more with less makes a case for a new approach to healthcare operations.

The healthcare industry can use prescriptive analytics to recommend

patients’ or providers’ best course of action. It can also compare multiple
“what if” scenarios. For example, if a healthcare provider is considering
adding a new service, prescriptive analytics can assess the impact of
choosing one service over another. With it, healthcare decision-makers can
optimize business outcomes and make more informed decisions about
patient care.

In a hypothetical example, a health insurer uses predictive analytics to spot

a pattern in its claims data. The previous year shows a significant portion
of its diabetic patient population also suffers from retinopathy. The insurer
then estimates the probability of an increase in ophthalmology claims
during the next plan year. Prescriptive analytics can model the cost impact
if average ophthalmology reimbursement rates increase, decrease or
remain the same for the following year. It can then recommend a course of
action. This type of analytics can be beneficial for insurance companies in
predicting trends and planning for the future.

The early adopters of big data analytics are doing great things for patient
care and their financial health. For example, predictive risk scores help
prevent suicides and increase watchfulness in the ICU unit. In addition,
Gene expression tests allow doctors to identify high-risk people and
provide new hope by discovering targeted treatments.

Cognitive computing engines, natural language processing, and free-text

analytics have the potential to help providers pinpoint diagnoses that might
otherwise elude them. These tools also provide population health
management capabilities for payers and healthcare professionals to
highlight those most at risk of being readmitted or developing costly
chronic diseases in their care populations.

Predictive analytics may be complicated, but healthcare organizations

across the country are using predictive analytics to improve patient care
and save lives. Data insights in Health care should also provide answers
and action options to questions such as:

- What’s the best way we can maximize outcomes for patients?

- How can we change processes and maximize interactions with
patients?
 - Which patient should healthcare providers intervene with, in order
not just to do patient management (once the patient gets sick) but
to do disease prevention (saving patients from getting sick
altogether)?

Another problem it should provide options to solve is how to manage

disease once the patient gets them, and figuring out how to get involved
early, such as the recent Covid-19 disease.

3. Operations. Prescriptive analysis supports decision-making relating

to manufacturing and logistics,
optimizing a company’s supply chain.
The main applications of prescriptive
analysis to the supply chain include:

> Consumer trend forecasting:

prescriptive analytics can make
deductions regarding consumers’ intentions and predict demand
patterns to make the right decisions. Calculating consumer trends
lets you establish optimal inventory levels to satisfy demand,
prevent stockouts and avoid overstock.

> Promotion of product traceability. Prescriptive knowledge

facilitates real-time information on products (traceability) at any
point across the supply chain, e.g., their location, transport
conditions and the logistics and manufacturing processes they’ve
been through.

> Greater control over supply chain information: prescriptive

analytics paves the way for real-time inventory management,
enabling organizations to issue instant supply orders and
accurately track their goods, among other advantages.

The McKinsey publication A more resilient supply chain from optimized

operations planning explains how prescriptive analytics can benefit
companies: “Optimisation in operations planning involves determining the
optimal choices for a set of decisions in a given business environment and
business target. This type of optimization generally works best with
prescriptive models that provide the ideal set of decisions as an output.”

In today’s changing, competitive market, consumer trends can lead to

supply chain disruption. Against this backdrop, prescriptive analytics has
become an ally, helping companies to stand out as leaders in their markets.
In the past, organizations used their know-how and experience to make
rough estimates — with varying degrees of success — on sales and the
product quantities needed to meet demand.

Nowadays, prescriptive analytics is a technological solution to anticipate

future logistics scenarios and make decisions based on data analysis.

Ecommerce giant Amazon recently

announced that it would be shipping
the products even before the
customer buys it. Strange, how can
Amazon know who would buy which
product at what time? Amazon uses
Predictive analytics blended with
descriptive analytics (trends, patterns, exceptions) of customers’ historical
shopping data to predict the probability of a customer to buy a product
with the date-time information.

The system prescribes Amazon to ship the product to the customer with an
expected order date. Amazon then packages and dispatches the product to
a local warehouse until a real order has arrived. As soon as the order arrives,
the product is delivered to the customer – almost instantaneously. Many
other players also use analytics for customer experience management
including Flipkart, eBay, makemytrip, BigRock, HomeShop18 etc.

Data insights in Operations should also provide answers and action

options to questions such as:

- How do we better anticipate demands at any given time?

- How do we maintain the quality of the product in relation to cost
given that the quantity of material and labor vary over time?
- How do we figure out ahead of time that a certain product is
likely to fail and fix the product before it does fail, so we can
maximize the uptime?
4. Human Resources. In HR, this is known as “Workforce Analytics.”
Workforce analytics refers to
gathering HR data, understanding
what it means within the context of
business goals and using it to
optimize decision-making and
operations.

Historically, this process required several teams and a lot of time. But
modern analytics technology has simplified an employer’s ability to create
and use meaningful data. Why are workforce analytics important?
Workforce analytics helps employers identify potential causes of
performance-related problems and address them in ways that maximize
opportunities and minimize risk. Without this data, employers may have
limited tools or no tools to support strategic decisions.

Consider employers that want to reduce the cost of overtime resulting from
unplanned absences. They can use advanced analytics to compare
overtime rates against absences and look for trends. They may also
conduct a root cause analysis to understand whether the unplanned
absences are connected to specific teams or supervisors. This information
can then be used to develop an appropriate corrective action. Manager and
employee training, for example, may help improve attendance and
scheduling management and engagement.

What are the benefits of workforce analytics? Workforce and HR analytics

that are easy to understand and act upon can potentially help businesses
achieve strategic objectives. Specifically, employers may be able to use
analytics to:

1. Make more timely and appropriate decisions. Leveraging internal

and external data may help employers anticipate what could
happen to their organization and quickly address or mitigate
potential risks.

2. Recruit talent and keep them engaged. With advanced analytics,

HR professionals can potentially identify the type of talent needed
for a particular job. The data can also help employers ensure that
employees remain motivated to fulfill the company’s mission.

3. Improve productivity. Centralized, real-time workforce metrics

 allow employers to monitor employee productivity and adjust
operations as needed.

4. Reduce costs. Employers that have access to benchmarks for

their industry and location may be able to recruit employees with
competitive market rate pay rates. Once hired, analytics can be
used to monitor scheduling and help control overtime costs.

5. Enhance data security. Workforce analytics that are cloud-based

typically have a flexible and resilient infrastructure designed to
help protect sensitive employee data and reduce risk.

Given these benefits, companies should ideally be eager to early-adopt and

explore how prescriptive analytics could transform their HR functions.
Analytics in HR should also ideally provide answers to the following
questions:

- How do we best reward our outstanding employees?

- How to keep the morale of the employees high even in
times of recession?
- What recruits should we hire?
- Shall we dole out overtime or not? If yes, how much overtime is
beneficial to the company and the employees?

5. Insurance and Risk. Data and analytics have driven pricing and
claims-related business decisions for
decades. Now with the proliferation of
big data, more and more insurance
companies are mining and using this
data to predict what will happen to
their businesses in the future—and to
drive decisions that will positively
affect these outcomes. In 2021, insurance companies invested $3.6 billion
in big data analytics. In return, they’ve seen a 30% increase in efficiency,
40–70% cost savings, and a 60% increase in fraud detection rates. The
insurance industry is a leader in the use of big data analytics and now uses
this information at almost every stage of its everyday operations.

While insurance analytics can improve the efficiency of claims, policy, and
sales processes, that’s just the tip of the iceberg. Other key benefits of data
analytics for insurance include:

1. Lead Generation. Analysis gives marketers and insurers a clear

understanding of their Customer Acquisition Cost or CAC,
displaying how much is spent on organic lead generation and
third-party vendors. The end result of this is that it puts out a clear
view of where companies should invest their resources.

2. Better Customer Satisfaction. Companies that accurately predict

their customers’ needs, whether it’s support or a particular
product, are more likely to exceed expectations and reap the
rewards.

3. Less Fraud. Claims fraud has always been one of the biggest
challenges facing the insurance industry. In the United States
alone, fraudulent claims cost at least $80 billion per year —an
estimated 10% of all payouts. Data analytics in insurance makes
fraud-detection processes faster and more accurate.

Analytics make it easier to spot trends while advanced analytics

and predictive modeling use historical data (e.g., past claims,
frequency of claims) and externally sourced information (e.g.,
credit scores) to flag claims with a high probability of being
fraudulent. These data analytics have become so advanced that
some predictive analysis models can even pinpoint customers
who are most likely to submit fraudulent claims, empowering
insurers to take preventative measures before they’re taken for a
ride.

4. Faster Underwriting. Data analytics for insurance is eliminating

intuition and guess work in underwriting by using predictive trend
data to produce comprehensive risk assessments. These
assessments reduce the time burden of manually assessing risk
profiles while improving underwriters’ ability to set premiums that
accurately reflect each policyholder’s level of risk.

5. Business Growth. The intelligent and ongoing use of data

analytics should culminate in business growth. Targeted
marketing messages, better customer satisfaction, less fraud, and
faster underwriting can only translate into improved revenue. By
 selling more effectively, cutting fraud-related costs, and improving
risk assessment, companies that choose to harness the power of
insurance data analytics improve their bottom line. Predictive
analytics, in particular, can help companies identify money pits
and mitigate them before they get out of hand.

From the pandemic to climate change, customers face heightened

uncertainty about their safety and well-being. They also question whether
their data will be used responsibly—but they are willing to share it in
exchange for value. The use of customer data to generate relevant, real-
time usage- and behavior-based offers that help customers mitigate,
manage, and recover from loss can help insurers build trust with
customers. That’s the value advanced data analytics can deliver both to the
insurance customer and to the insurer.

Data Analytics in Insurance and Risk should also ideally provide answers
to the following questions:

- Can we detect fraud and non-compliance?

- Are these customers good credit risks?
- Can I forecast expected revenue?
- Can I better price my insurance products?

6. Oil Industry through Fracking. In the past years, fracking has taken
an enormous flight, especially in the
United States. In 2013 alone, $ 31
billion was spent on suboptimal frack
stages across 26,100 U.S. wells. In
order to know where to frack, make the
process safer and to optimize the
fracking process, massive data sets
(up to petabytes of data) are required.
Data sets such as sounds (of fracking and drilling), images (seismic, well
logs), videos (cameras monitoring the fracking and sensors measuring all
kinds of variables), text documents (notes by drillers) and other kinds of
data have to be analyzed in real-time to recommend the most optimal
fracking location and process in order to have the best result.

Prescriptive analytics substantially reduces the financial risk of fracking by

revealing key insights, modeling potential drilling scenarios, and identifying
areas of opportunity. It reduces the amount of drilling needed to find
productive shale oil, which means less harm to the environment. Statistics
show that fracking only recovers about 20 percent of the oil trapped in shale
rocks.

By assimilating vast amounts of different types of data—well log images,

seismic readings, video image feeds, data from field sensors, drilling notes,
and so on—big data techniques can assimilate that data into an integrated
profile. That data can then be extrapolated to determine what is going to
happen where and when. In oil exploration, that means determining where
the best place is to drill.

To accurately interpret seismic images, prescriptive analytics needs to

combine data from different computing and scientific sources. For example,
by combining machine learning, pattern recognition, and computer vision,
algorithms can produce images to reveal where and how to drill. The “where
and how” to drill is the prescriptive component. Beyond oil exploration,
prescriptive analytics can model well performance and help companies
understand where they’ll get a certain amount of oil. Digitizing well logs
creates depositional subsurface maps. This involves using data from oil
production, subsurface completion, and other sources.

Prescriptive analytics also can model limitations related to fracking

exploration. For example, a shortage of pressure pumping equipment and
available labor can be modeled to assess the costs of drilling and
production against return on investment. Prescriptive modeling is ideal for
decision mapping, balancing current business demands and business
objectives against constraints.

Prescriptive analytics is also making oil production more efficient. For

example, with many fracking operations only a small portion of available
shale oil is actually extracted. Using prescriptive analytics, you can make
improvements that will make the next well more productive and
cost-effective. For example, many shale producers are drilling unfinished
wells because they have to drill to keep their lease, and a new lease could
cost as much as five times more. Drilling is only 30 percent of well
completion costs where fracking is 70 percent, so prescriptive analytics can
be used to model where to drill and when to maintain potentially productive
land leases.
More importantly, you can watch for unsafe drilling behavior and prevent
potentially fatal accidents.

7. Travel and Transportation Optimization. The travel industry is an

industry that sees a lot of potential in
the latest addition of analytics. Online
travel websites, such as airline
ticketing services, hotel websites or
car rental websites, have turned to
prescriptive analytics to sift through
multiple complex iterations of travel
factors, purchase and customer variables such as demographics and
sociographics, demand levels and other related data sources to optimize
their pricing and sales.

Other applications in the travel industry are segmenting (potential)

customers based on multiple data sets to understand how to spend the
marketing budget. More accurate targeting of course results in higher
conversion rates and to be able to make detailed segments, a lot of different
variables are required. The InterContinental Hotel Group for example uses
650 variables to determine the best price/product combination for the right
customer.

Another use case of prescriptive analytics is route optimization for the

logistics industry. UPS is the best example of this. By analyzing and
combining hundreds of data sources, UPS can push 10.000s route
optimizations per minute to all of their trucks. This saves the company
millions of dollars on fuel a year.

UPS uses a wide variety of data to

reduce the environmental impact of
their business. They combine data
from engines that provide UPS with
insights in performance and condition
of their vehicles. Apart from the engine
performance, also speed, number of
stops, mileage and miles per gallon etc. are measured. They use GPS data
that captures driver behavior and safety habits. Sensors in vehicles report
data on emissions and fuel consumptions.
All kinds of devices monitor deliveries and customer services as well as
address points and routes traveled that can be analyzed to optimize the
routing again. In addition, all drivers have state-of-the-art handheld devices
that also collect a lot of data. Long story short: they collect data at every
possible moment.

Because of all the data they collect, UPS is capable of creating unique
services for its customers. One of such services is My Choice, which allows
customers to adjust delivery timing and location via the smartphone at any
moment. But they also invest a lot of money to speed-up the delivery
process. As such they have connected their systems with systems of
millions of their customers. This allows them to move the data ahead of
packages and, for example, pre-clear packages before they arrive at US
customs. As such, international packages can be moved directly to
domestic packages and be delivered the following day.

Their latest project is called ORION and costs $ 1 billion a year. ORION
stands for On-Road Integrated Optimization and Navigation. ORION uses
fleet telematics and advanced algorithms to take route optimization to the
next level. Started in 2013 and deployed across North American routes in
2017, ORION is viewed as the largest operations research project in the
world. With 250 million+ data points, ORION is capable of delivering tens of
thousands route optimizations per minute based on real-time information.
Next to that, the system allows UPS managers to monitor driving habits of
its workers and see when a driver backs up a truck too often or makes too
many U-turns. This information could signal drivers that require additional
training to become more efficient.

8. Weather Forecasting. The amount of available data can be both a

blessing and a curse. While it’s
important to have extensive data from
many sources to gather accurate
information and offer pointed
predictions, it can be complicated to
select which weather models are most
applicable to the data at hand. In
addition, the further into the future a forecast is made, such as a week out
or even longer, variables such as the constantly changing atmosphere can
complicate matters. This is why continuous, real-time analysis plays a
crucial role in accurate weather forecasting.
By incorporating tools from artificial intelligence and machine learning
algorithms, it’s possible to significantly increase the reliability of weather
forecasts. Weather models can be made much more accurate by using
these tools not only to quickly process huge amounts of weather data but
to offer a more refined approach the more the models are used.

Data Analysts who are trained to predict the weather offer many benefits to
various stakeholders, from individuals to entire regions of a country.
Weather Analysts rely on various data analytic and predictive analytic
techniques in order to provide the most accurate and helpful forecasting
models possible. Accurate weather predictions have many real-world
ramifications; they can help people decide on the best course of action to
protect themselves, their products, and their belongings. By predicting what
the weather will bring, lives can be saved.