Predictive Analytics for Business Forecasting

The past few decades have witnessed growing adaptability of analytical forecasting tools in the corporate sector. This may be attributed to increase in the complexity, competitiveness and the rate of change in the business environment. The objective of this programme is to present a comprehensive view of the various tools and techniques used in forecasting for managerial decision making including the problem of demand estimation, market size determination, sales projections, analyzing and predicting stock prices. The methodology, covering various time series analysis techniques, as well as regression methods, will be presented with appropriate mix of case analysis and numerical demonstration with the aid of software packages so as to enable the participants to meet their own forecasting needs.

Participants would be split into small groups to work on forecasting projects that would be either decided/brought by the participants or would be given to them. Learning from these projects will be the key take-away from the program. Better part of a program day is reserved for participants to work on this project. In addition, the participants are expected to work during the evenings of the program days.

R is a free open source software which will be used throughout the course. Prior experience in working with R will be beneficial. On joining the program, introductory guidelines will be sent to the participants about a week prior to the program to help them getting familiarize with R. In addition, help would be available during the program days to pick up necessary coding with R.

CONTENT

• Basic statistical concepts: standard error, confidence interval estimation, significance values in testing,
• Simple and Multiple regression
• Logit Probit Models
• Time Series Decomposition Models
• Smoothing Models
• Box Jenkins (ARIMA)
• ARIMA with regression errors and ARIMAX
• Models for timeseries with multiple level seasonality.
• Bass Model for new product forecasting
• Combing Forecast and Forecast Evaluation
• Project work and presentations
• Forecasting projects from Industry

KEY BENEFITS/TAKEAWAYS

On completion of this course, the participants should beable to

• understand and implement various time series (Decomposition, smoothing, Box Jenkins) and regression models to come up with objective forecasts of sales, demands, stock prices etc
• produce and interpret forecasts of qualitative response based on probit and logit models
• forecast growth of a new product using Bass-type models
• critically compare and combine various methods used for forecasting.