|
Chapter 9: Predictive Analytics
We are now aware of how data and the analysis of it are vital for a company
to be able to function optimally. We’ll now examine another branch of data
mining that can help grow a company. In this chapter, we’ll be taking a look
at predictive analytics, starting with what it is and how it can help a
company function more efficiently.
Defining Predictive Analytics
Predictive analytics is used to make predictions about unknown future
events. It uses many techniques, such as statistical algorithms, data mining,
statistics, modeling, machine learning and artificial intelligence, to analyze
current data and make predictions about the future. It aims to identify the
likelihood of future outcomes based on the available historical data. The
goal is therefore to go beyond what has happened to provide the best
assessment of what will happen.
More and more companies are beginning to use predictive analytics to gain
an advantage over their competitors. As economic conditions worsen, it
provides a way of getting an edge. Predictive analysis has become more
accessible today for smaller companies, healthcare facilities, or smaller,
low-budget organizations. This is because the volume of easily available
data has grown hugely, computing has become more powerful and more
affordable, and the software has become simpler to use. Therefore, one
doesn’t need to be a mathematician to be able to take advantage of the
available technologies.
Predictive analysis is extremely useful for a number of reasons: Firstly, it
can help predict fraud and other criminal activity in places from businesses
to government departments. Secondly, it can help companies optimize
marketing by monitoring the responses and buying trends of customers.
Thirdly, it can help businesses and organizations improve their way of
managing resources by predicting busy times and ensuring that stock and
staff are available during those times. For instance, hospitals can predict
when their busy times of the year are likely to be, and ensure there will be
enough doctors and medicines available over that time.
Thus, overall efficiency can be increased for whatever organization utilizes
the data effectively.
Different Kinds of Predictive Analytics
Predictive analytics can also be called predictive modeling. Basically, it is a
way of matching data with predictive models and defining a likely outcome.
Let’s examine three models of predictive analytics:
Predictive Models
Predictive models are representations of the relationship between how a
member of a sample performs and some of the known characteristics of the
sample. The aim is to assess how likely a similar member from another
sample is to behave in the same manner.
This model is used a lot in marketing. It helps identify implied patterns
which indicate customers’ preferences. This model can even perform
calculations at the exact time that a customer performs a transaction.
The model can be used at crime scenes to predict who the suspects may be,
based on data collected at the scene. Data can even be collected without
investigators having to tamper with the crime scene, by use of, for example,
3D laser scanners, which make crime scene reconstruction easier and faster.
The investigators don’t even have to be at the actual scene, but can examine
it from their office or home. Nothing at the actual crime scene is disturbed,
and all the images and other data are stored for future reference. The scene
can later be reconstructed in a courtroom as evidence.
Descriptive Modeling
Descriptive modeling describes events and the relationship between the
factors that have caused them. It’s used by organizations to help them target
their marketing and advertising attempts.
In descriptive modeling, customers are grouped according to several
factors, for example, their purchasing behavior. Statistics then show where
the groups are similar or different. Attention is then focused on the most
active customers. Customers are actually given a “value”, based on how
much they use the products or services and on their buying patterns.
Descriptive modeling finds ways to take advantage of factors that drive
customers to purchase.
It’s worth bearing in mind that while descriptive modeling can help a
business to understand its customers, it still needs predictive modeling to
help bring about desired results.
Decision Modeling
The rapidly growing popularity of decision models has enjoyed much
attention recently. Modeling combines huge quantities of data and complex
algorithms to improve corporate performance. Decision models can be
extremely useful, helping managers to make accurate predictions and
guiding them through difficult decisions, unhindered by bias and human
judgement alone. By using models, data can be used more objectively.
Managers need to be able to use data to make decisions. A decision-
centered approach is quickly becoming the central analytics focus for most
businesses. The ability to model and find solutions for complex issues
allows for better decision making.
Decision models can help with such problems as to how to optimize online
advertising on websites, how to build a portfolio of stocks to get maximum
profit with minimum risk, or how online retailers can deliver products to
customers more cheaply and quickly. It can also help product developers
decide which new products to develop in the light of market trends.
When at the decision-making stage, it’s best to focus on action-oriented
decisions that are repeatable. Decisions should be based solidly on the
available data and be non-trivial, and should also have a measurable
business impact.
A good model will have well-defined questions and possible answers. It
will be able to be easily shared among team members.
While managers should use models to the maximum, they should also bear
their limitations in mind. For instance, models can predict how many days
of sunshine and rain a farming operation in a certain area may receive that
year, but they cannot actually influence the weather. It may predict the
likely amount that customers may spend on a product in a given year, but it
will never be able to directly control their spending habits.
Do'stlaringiz bilan baham: |
