Data Analytics with R, Excel & Tableau

An Introduction to Predictive analytics: Using data science to predict the future

Today most of the fortune 500 companies are using the power of predictive analytics to make better business decisions.  According to one research, the market for predictive analytics will be almost $11 billion by 2022. This is HUGE!  And with predictive analytics yet to reach its peak, the market is only going to grow from here. This means there is a very high requirement for Predictive Analytics experts, and more of such experts will be needed in the industry in coming future.  Lets understand in detail what predictive analytics is all about.

What is predictive analytics ?

Predictive analytics is a branch of data science and statistics understanding the patterns from the historical data and applying these patterns to current data to understand the future possible outcomes. Today organisations are sitting on tons of data.  For organisations, it makes perfect sense to analyse this existing historical data for trends, build a model based on these patterns and apply the model to the current data to predict the future events.  Tools required for predictive analytics can include data mining techniques, artificial intelligence, statistics, machine learning.

Big data predictive analytics, with the power of big data applied through predictive models has become all the more important tools for the organisations today.

How to perform predictive analytics ?


Steps for performing predictive analytics are as follows :

Understand the objectives and define the expected results

The first and foremost step is to clearly understand and define the objectives of predictive analytics that is to be performed. This requires good understanding of the capabilities of predictive analytics, kind of data that organisation generates and good understanding of the business along with its key metrics.

Capture the data

Once the objectives of the project are clear then next is to capture the relevant historical data.  Data may be collected from different sources available with the organisation.  Data will then need to be cleaned. Data from different sources will required to be combined into datasets.

Build the predictive model

Using statistical analysis and data mining we would now be required to recognize the patterns and trends in the historical data. Based on these patterns a predictive model is to be build.


Apply this predictive model to the current data to predict the future outcomes.


The most important step – act on the prediction.

Constant review of the prediction model is required, so that it keeps getting robust with the time.

Analytical languages like R, Python, Scala are used more often for building complex predictive models. Organisations have also been using enterprise tools provided by companies like Microsoft, IBM, SAS for their predictive analytics requirement.

Types of predictive analytics

Predictive analytics is sometimes used as a blanket term for three types of analytics :

  • Descriptive analytics is about capturing and analyzing past historical data to understand the past patterns
  • Predictive analytics goes further and with analysis of past data come with a data model which can predict possibility of future events
  • Prescriptive analytics is the stream of data science which uses past patters to not just come with future possible trends but also suggest business decision options

Predictive analytics techniques

There are multiple techniques that can be used to build a predictive model. Choice of the technique depends on the complexity involved, type of existing data and nature of the problem. Some of the popular techniques are as below :


Regression techniques are the most important tools for predictive analytics. Whole idea of a regression model is to find a mathematical pattern between two or more different variables by performing statistical analysis on the past available data. Most commonly used type of regression technique is linear regression model.

Time Series models


Time series model
Time series model

This technique is used for forecasting the future value of variables involved based on the historical data taken over time. This is important especially when relation between variables is time dependent – e.g. seasonality. Market demand for a product may depend on the season or may be cyclical. In such scenarios time series is the better technique to be used.

Machine Learning


Machine Learning
Machine Learning

Machine learning includes sophisticated statistics and regression models. Best part of machine learning techniques such as neural networks is the ability is the ability to learn about the relationship of different variables through human like cognition and through experimentation and training. For this reason machine learning is very helpful when relation between variables isn’t known or if the relation is very complex.

Some examples of its usage

Predictive analytics today find lot of usage in many fields including healthcare, travel, insurance, marketing, Fintech.  Some of the most popular usage examples of predictive analytics are :

Predicting Customer behaviour : Predictive analytics can be utilized to predict the behaviour patterns of individual customers and using that design the custom marketing campaigns to provide more value to these customers, help to retain them and increase their life time value from such customers. Online e-commerce companies use predictive analytics heavily.

Fraud Detection: Among the most popular usage of it is for the financial companies.

Fraud Detection
Fraud Detection

Organisations can detect the patterns of past frauds and anomalies. Based on the past patterns organisations can create models to predict future frauds through current data stream. This enables the organisation to take preventive actions to thwart off the fraud attempts and to minimize business risk.

Optimizing operations : Companies can use the predictive analytics to improve their supply chain operations. Organisations can predict the requirement of raw materials in advance, forecast the inventory, predict the demand and thus optimise better.


Predictive analytics has brought revolutionary changes in some industries like ecommerce, retail, Fintech etc. This is only a start. With the evolution of machine learning and AI – power of predictive analytics is expected to grow exponentially.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Related Articles