Big Data

BUSN 20800 – Spring 2021

Last updated: 2022-10-11 14:06:21. Homework answers available on request.

• All Lectures in Order
• All Homeworks in Order
• Final, pdf, Rmd

Support or Contact

Questions? Email me (cdowd@chicagobooth.edu) and I’ll help you out.

Before Class

• Syllabus.
• Textbooks:
  • “Elements of Statistical Learning” (ESL) by Hastie, Tibshirani, and Friedman – online and here
  • “Statistical Consequences of Fat Tails” by NN Taleb – online and here
  • “Causal Inference: The Mixtape” by Scott Cunningham – online
• HW0: An ungraded assignment for making sure everyone is up to speed. Entirely optional.
  • HW0 and the .Rmd code it is based on
  • HW0 Answers and .Rmd file

Week 1

• Lecture 1 – Admin, Syllabus, Stats 101, FDR Control – .Rmd file and PDF
• HW1a – Prediction Competition.
  • .Rmd
  • My prediction: code and data
• Lecture 2 – Predictions, FDR Control, and Loss – .Rmd and PDF
  • Lipids example:
    • Code.
    • Data is on canvas under lecture 2. “jointGwasMc_LDL.txt”
  • Diabetes Example:
    • Code: build pvals and do fdr
    • Data: base data and pvals
  • Simulations in lecture: Code
• Homework 1: FDR control
  • .Rmd
  • Data (not CSVs): reviews, products, word freq, words
  • Due April 6 before class.
  • Check out the code under Lecture 2 if you need assistance.
• Textbook references:
  • False discovery rate: Ch. 18.7 of ESL
  • Loss: Ch. 2.4, Ch. 7.2, and Ch. 10.6 of ESL

Week 2

• Lecture 3 – Regression: Linear and Logit – Rmd and PDF
  • OJ example: code and data
  • Spam example code and data
  • Tidyverse Intro: A brief introduction to tidyverse functions. Rmd (likely to be updated over the next few days)
• Homework 2 – Due Wednesday April 14
  • Rmd
  • Data and codebook
• Lecture 4 – Deviance, OOS, Bootstrap – Rmd and PDF
• Textbook References:
  • Linear Regression: ESL Ch. 3.2
  • Logistic Regression: ESL Ch. 4.4
  • Deviance: ESL p. 124
  • Out-of-Sample: ESL Ch. 7.1 (uses the term ‘generalization error’ or ‘validation error’)
  • Bootstrap: ESL Ch 7.49

Week 3

• Lecture 5 – Variable Selection – Rmd and PDF
  • For those of you struggling with homeworks, there is a lot of valuable code in the old Lecture Rmd files
  • Semiconductors: Data, code in lecture Rmd
  • Comscore: Code in lecture Rmd, Data: domains, sites, total spend
• Lecture 6 – Cross Validation – Rmd and PDF
  • The last few slides include a lot of useful code. If you can interpret and run the cross-validation code there, you should gain a solid grasp on cross-validation generally.
• Homework 3 and Rmd
  • Due Next wednesday at midnight.
  • Some of the ratios etc may be inverted in the solutions. This shouldn’t change any interpretations.
• Textbook References
  • AIC: ESL Ch 7.5
  • BIC: ESL Ch 7.7
  • Stepwise: ESL Ch. 3.3
  • LASSO/Ridge/Shrinkage: ESL Ch. 3.4, 3.6, 3.8
  • Cross-Validation: ESL Ch 7.10
  • Bias-Variance Decomp: ESL Ch 7.2, 7.3

Week 4

• Lecture 7 – Classification – Rmd, and PDF
  • Credit Data and codebook
• Homework 4 and Rmd
• Lecture 8 – Classification 2 – Rmd
• Surveys: General Feedback and 1-on-1 Meeting Signup
• Textbook References:
  • KNN: ESL Ch. 2.3.2, 13.3
  • Binomial Classification: ESL Ch 4, Ch 9.2.3, Ch 13.3
  • Sensitivity/Specificity/ROC curve: Ch. 9.2.5
  • Multinomial Regression: Ch. 4.4.1 (not the most helpful)

Week 5

• Lecture 9 – ROC and Trees – Rmd
• Lecture 10 – Trees and Forests – and Rmd
• Textbook References:
  • Trees: ESL Ch. 9.2
  • Bagging: ESL Ch 8.7
  • Forests: ESL Ch 15

Week 6

• Lecture 11 – Forests and Boosting – Rmd
• Predictions Competition 2 and Rmd
  • My predictions, Rmd, and data
• Lecture 12 – Boosting, Ensembles, and Rolling Block CV – Rmd
• Homework 5 and .Rmd
• Textbook References:
  • Boosting: ESL Ch. 10
  • Model Averaging: ESL Ch 8
  • Ensembles: ESL Ch 16

Week 7

• Lecture 13 – Data Cleaning – Rmd
  • Data: cigarettes
• Lecture 14 – Data Cleaning and Bayes – Rmd
  • Cleaned Zillow Data
• Homework 6 and Rmd – Cleaning
  • The textbooks below may help.
• Textbook References: (giant references for R) Both free online.
  • R for Data Science – Hadley Wickham: The wrangle section particularly.
  • R Cookbook: Help with various data types

Week 8

• Lecture 15 – Causal Inference 1: RCTs – Rmd
  • JTPA Data: raw, cleaned, cleaning code
• Optional Predictions 3 – Sunday Cases Redux – Rmd
  • My predictions, Rmd, and data
• Lecture 16 – Causal Inference 2: Targeting, Observational Methods – Rmd
• HW7 and Rmd
• Textbook References:
  • RCTs/ATEs: Causal Inference Mixtape Ch. 4
  • Observational Causal: Causal Inference Mixtape, Chapters 6, 7, 8, and 10 for Regression Discontinuity, Instrumental Variables, Diff-in-Diff, and Synthetic Controls respectively
  • Targeting/Heterogenous Treatment Effects: (no good texts at the moment, so I’ve linked a few papers) Tibshirani, Hastie, et al (and here). CATE meets ML. Optimal Targeting with Heterogenous TEs

Week 9

• Lecture 17 – Neural Nets, SGD, Optimization – Rmd
• Lecture 18 – Review of Course – Rmd
• Textbook refs: Neural Nets – ESL Ch. 11