96799 - FORECASTING AND PREDICTIVE ANALYTICS

Learning outcomes

At the end of the course the student has a wide knowledge of the most important statistical techniques employed for forecasting and prediction purposes in modern business activities. In particular the student is able to: - Select the most appropriate predictive model to solve the business problem at hand; - Analyze the data and perform predictions using the statistical software R; - Report the results in a proper format for the business management.

Course contents

1. A probabilistic approach to the business prediction and forecasting problem
   1. Characterize the business variable of interest using a probabilistic model
   2. Define predictions and forecasts in terms of conditional quantities of the probabilistic model
   3. Distinction between the prediction/forecasting problem and the identification of causal effects
2. Data structures and the distinction between predictions and forecasts
3. Evaluation of predictions and forecasts
   1. Distinction between training and test samples
   2. Identification of an appropriate criterion to evaluate predictions/forecasts
   3. Model selection based on prediction/forecast evaluation
4. Linear predictors
   1. Least squares regression, CEF and normal regression
   2. Constrained CEF and generalized linear models
   3. Quantile regression and parameter regression
5. Non-linear predictors
   1. Generalized additive models
   2. Generalized random forests
6. Forecasting models
   1. Time series predictions
   2. Autoregressive models
   3. Trend, seasonality and non-stationarity
   4. Time-varying parameters

Readings/Bibliography

Suggested textbook:

Diebold, F. (2017). Forecasting in Economics, Business, Finance and Practice. Open Textbooks.

Hansen, B. (2021). Econometrics. Open Textbooks.

James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An introduction to statistical learning. Springer.

Kuhn, M., & Johnson, K. (2013). Applied predictive modeling. Springer.

Schumway, R. & Stoffer, D. (2019). Time series: a data analysis approach using R. CRC Press

Stock, J. H., & Watson, M. W. (2015). Introduction to econometrics 3rd ed. Pearson

Teaching methods

• Presentation of the main theoretical aspects concerning the methods introduced in the course.
• Presentation of R packages implementing the methods.
• Presentation of business case studies and discussion of appropriate methods to solve the problems.

Assessment methods

Students will have to submit a final project and discuss it in an oral examination. The project can be developed by groups of maximum 3 students.

The topic of the project needs to be agreed with the instructor at the beginning of the course. The submitted project must contain: problem statement and final objective; data description and exploratory data analysis; discussion of the methods selected; empirical analysis and discussion of the results; easily readable R code used to perform the analysis.

Evaluation of the final project will take into account the difficulties posed by the problem and the data analysis and the group composition. The final grade will weigh the project evaluation and the oral discussion.

Teaching tools

• Slides
• Datasets
• R Scripts

Office hours

See the website of Luca Trapin