www.apress.com

16/07/2018

Revolutionizing the Art of Business: Machine Learning for Decision Making

By Deepti Gupta

Data science and machine learning are not just buzz words of the moment but have rather transformed how business is conducted in today’s world. Incorporation of technology and digitization of almost every feasible aspect of business provides access to abundance of data. This data can provide insights which offer a scientific basis for decision making. 

To keep up with constantly changing trends and to have ability to self-correct based of most recent information necessitates the machines with analytical capabilities. Machine learning is a data analysis method used to automate analytical model building algorithms.  Essentially, it makes the computer to automatically learn from the data and make the decisions and predictions based on the past data without being explicitly programmed.

Machine learning applications can be seen across the industries from fraud detection to customer engagement, churn, cross-sell & up-sell, predicting readmission rate, recommendation engines, text-to-speech and many more. In this blog, I have introduced different types of machine learning techniques and their applications in different business sectors. Machine learning techniques can be broadly classified into two categories: supervised learning and unsupervised learning. Figure 1 displays the Machine Learning models.


Figure 1: Machine learning model

Supervised learning models are built using labeled data. This type of data consists of one or more input variables (X) and a corresponding output variable (Y). The objective of these models is to predict for output variable (Y) on the basis of new input data(X). The goal of supervised machine learning algorithm is to learn a function which correlates the input variables to output. In a very simple form can be expressed as:

                                     Y=f(X)

WhereY: is output or target variable, X: is an input or explanatory variable 

And f( )  is function of which correlates X to Y.

Supervised learning is mostly used in context of classification and regression problems. For example banking industry would like to predict the fraudulent activities, telecom industry would like to predict customer churn, healthcare industry would like to predict the malignant and benign breast cancer, airline industry would like to predict ticket prices and many more. All these real time industrial problems can be solved by using common supervised learning algorithms like logistic regression, decision tree, random forest, neural network and linear regression.

On the other hand in unsupervised learning there is no labeled output, so the objective in unsupervised learning is to find the pattern and trend within a set of data points. Unsupervised learning is mostly used in context of clustering and association analysis.  For example in Fast Moving Consumer Goods (FMCG) industry for their marketing campaigns the analyst is trying to segment the consumers into best, prospect and lost customers by analyzing the past data, in such scenario unsupervised cluster analysis would be a great starting point for the customer segmentation analysis. Common unsupervised learning algorithms are k-means clustering, DBSCAN, hierarchical clustering, principal component analysis, singular value decomposition and many more.

Data scientist use various types of supervised and unsupervised machine learning algorithms in order to find the hidden insights from the data that lead to the actionable insights. To decide which algorithm would be used to approach the specific business problem typically depends upon the target application and factors related to the attributes and volume of the data on hand. Most industries with high volume of data have recognized the importance of the machine learning techniques and are applying it on real time in order to  strategies business decision more effectively and wisely or gain an advantage over the competitors.

About the Author

Deepti Gupta completed her MBA in Finance and PGPM in operation research in 2010. She has worked with KPMG and IBM private limited as Data Scientist and is currently working as a data science freelancer. Deepti has extensive experience in predictive modeling and machine learning with an expertise in SAS and R. Deepti has developed data science courses, delivered data science trainings, and conducted workshops for both corporate and academic institutions. She has written multiple blogs and white papers. Deepti has a passion for mentoring budding data scientists.

This blog post was contributed by Deepti Gupta, the author of "Applied Analytics through Case Studies Using SAS and R: Implementing Predictive Models and Machine Learning Techniques".