BUAL 6610/6616-Predictive Modeling II  

Spring 2018
Dr. Pei Xu
Harbert College of Business
Auburn University

Course Objective

This course aims to go beyond the traditional regression method and to provide a much applied overview to those modern predictive methods, such as Neural Network and Deep Learning, Random Forest, Boosting and Bagging. We will cover these approaches in the content of Marketing, Finance and other important business decisions. Those approaches are used to extract business intelligence by leveraging firm's business data along with online social media content for various applications, including social media analytics, market analysis, credit risk assessment, and text and web mining.


Final Project


Unit 1: Course Introduction (Week 1)

1. Syllabus; 2. Introduction to Business Analytics

Useful Links:



Unit 2: Accessing and Assaying Prepared Data (Week 2)

1. Slides: PredictiveModelingOverview

2. Lab: Introduction to Python (notebooks); Introduction to Pandas (ipynb)

3. Extra reading: Big Data_New tricks for Econometrics; Theoretical Impediments to Machine Learning.pdf


Unit 3: Evaluation of Binary Classifier, ROC (Week 3)

1. Slides: Binary Classifier & ROC <pdf, ipython>

2. Python Lab: Stock performance dataset <descriptive information: pdf, ipython>

Evaluate Model Performance; calculate prediction accuracy, sensitivity, specificity, ROC.

3. Extra Reading: Logistics Regression; ROC


Unit 4: Resampling Methods: Cross Validation & Bootstrap  (Week 4)

1. Slides: (1) Bias-Variance Tradeoff <slides>   

               (2) Resampling methods: Cross Validation & Bootstrap <pdf, ppt>

2. Python Lab: Use different Resampling Methods to Split the Smarket dataset; evaluate the model performance.

Holdout Validation; K-fold Validation

3. Reading: Recommended textbook: An Introduction to Statistical Learning (Chapter 2)


Unit 5: Decision Tree and Random Forest (Week 5)

1. Slides: Introduction to Decision Tree and Random Forest <pdf>

2. Python lab: Decision Tree notebook <Decision Tree.ipynb; pdf> <datafile: Hitters>; Random Forest notebook <ipynb>

3. Extra Reading: O'Reilly Data Science Salary Survey (2016; 2017); Predict Customer Retention

Case study: How a Japanese cucumber farmer is using deep learning


Kaggle Competitions

Challenge yourself with real-world machine learning problems: First Kaggle Project.ipynb


Unit 6: Support Vector Machine

Slides: 6.Support Vector Machines.pptx

Python notebook: Support Vector Machines.ipynb



Unit 7: Neural Network and Deep Learning

Concept, review questions <ppt>

Case study - Classification neural network(FIFA) ; Neural Network 2017


Unit 8: Text Mining

1. Slides: Text Mining-updated.pptx

2. Case: Facebook & Sales; Facebook & Sales-supplement.pdf; DirectTV

3. Lab: AmazonReview.zip

4. SASTextMiner-Tutorial.pdf



Unit 9: Social Media Analytics

 A quick puzzle -baseball cap.docx



Final project presentation

Cases: Target and Twitter cases.docx


Unit 10: Case study

AUFootball Facebook Comments <data, ipython notebook>