Business Analytics and Intelligence
Module Structure
Module 1: Foundation of Data Science
The process of fact-based decision making requires managers to know how to summarize, analyse, conduct hypothesis tests, interpret and communicate data using data visualization and descriptive statistics techniques to facilitate decision making. Statistical analysis is a fundamental method of quantitative reasoning that is extensively used for decision making. This module is aimed at providing participants with the most often used methods of statistical analysis along with appropriate statistical tests. The module is oriented towards application without compromising the theoretical aspects.Â
Module 2: Data Preparation and Imputation
Quality of the data is important for success of any analytics project.
Anecdotal evidence suggests that more than 80% of time taken for an analytics project is spent on data preparation and data imputation. In this short module, we will be discussing data preparation and imputation techniques before advanced analytics tools can be applied.
Module 3: Data Preparation and Imputation
Predictive analytics model predicts occurrence of future events such as demand for a product, revenue forecast, customer churn, employee attrition, fraud, default in loan repayment, etc. based on historical data. In many business problems, we try to deal with data on several variables, sometimes more than the number of observations. Regression models help us understand the relationships among these variables and how the relationships can be exploited to make decisions using supervised learning algorithms. Primary objective of this module is to understand how regression and causal forecasting models can be used to analyse real-life business problems such as prediction, classification and discrete choice problems. The focus will be case-based practical problem-solving using predictive analytics techniques to interpret model outputs. The participants will be exposed to software tools such as MS Excel, R, Python, SPSS and how to use these software tools to perform regression, logistic regression and forecasting.
Predictive Analytics Module Contents
Regression model building framework : Problem definition, Data pre-processing; model building; Diagnostics and Validation
Simple linear regression : Coefficient of determination, Significance tests for predictor variables, Residual analysis, Confidence and Prediction intervals
Multiple linear regression : Coefficient of multiple coefficient of determination, Interpretation of regression coefficients, Categorical variables, heteroscedasticity, Multi-collinearity, outliers, Autoregression and Transformation of variables, Regression Model Building
Logistic and Multinomial Regression : Logistic function, Estimation of probability using logistic regression, Deviance, Wald Test, Hosmer Lemshow Test, Classification table, Gini co-efficient; Multinomial logistic regression.
Forecasting : Moving average, Exponential smoothing, Casual models
Application of predictive analytics in retail, direct marketing, health care, financial services, insurance, supply chain, etc.
Module 4: Prescriptive Analytics
Optimization models are core tools used in prescriptive analytics and are used in arriving at optimal or near optimal decisions for a given set of managerial objectives under various constraints. Optimization techniques such as gradient decent plays an important role in many machine learning algorithms. Optimization is an integral part of operations analytics with specific applications in operations and supply chain management. The objective of the module is to acquaint participants with the construction of mathematical models for managerial decision situations and use freely available Excel Solver to obtain solutions and interpret the results.
Optimization Analytics Module Contents
Introduction to Operations Research (OR), linear programming (LP), formulating decision problems using linear programming, interpreting the results and sensitivity analysis. Concepts of shadow price and reduced cost.
Multi-period LP models. Applications of linear programming in product mix, blending, cutting stock, transportation, transshipment, assignment, scheduling, planning and revenue management problems. Network models and project planning.
Integer Programming (IP) problems, mixed-integer and zero-one programming. Applications of IP in capital budgeting, location decisions, contracts.
Multi-criteria decision making (MCDM) techniques: Goal Programming (GP) and analytic hierarchy process (AHP) and applications of GP and AHP in solving problems with multiple objectives.
Non-linear programming, portfolio theory, gradient decent technique.Module 4: Prescriptive Analytics
Module 5: Stochastic Models
Stochastic models offer a powerful analytical approach to model and examine complex problems in the domains of finance, retail, marketing, operations and economics under uncertainty. In management as well as in business, many measurements change with time and are inherently random in nature. Stochastic models can be used to model and measure changes in metrics used for finance, marketing, operations, supply chain, etc. over a period of time. The objective of this module is to provide an introduction to stochastic processes and their applications to business and management. Stochastic models are also the basis for reinforcement learning algorithms.
Our approach will be non-measure theoretic, with an emphasis on the applications of stochastic process models using case studies.
Stochastic Models Module Contents
Introduction to stochastic models, Markov models, Classification of states, Steady-state probability estimation, Brand switching and loyalty modelling, Market share estimation in the short and long run. Google's ranking algorithm.
Poisson process, Cumulative Poisson process, Applications of Poisson and cumulative Poisson in operations, marketing and insurance. Measuring effectiveness of retail promotions, warranty analytics.
Monte Carlo simulation.
Reinforcement Learning Algorithms : Dynamic Programming; Markov decision process, Applications of Markov decision process in sequential decision making.