This is a repository to organize the teaching material for a new course at Monash University, to be jointly taught by Di Cook, Rob Hyndman and Souhaib Ben Taieb.

M 1-2 E365; T 4-6 E157; Th 11-12 H7

• What is statistical learning?
• Types of models: regression, classification, unsupervised
• Model estimation: numerical optimization
• Model assessment and selection: predictive accuracy
• Resampling methods: bootstrap, permutation, cross-validation, bagging
• Regression: splines, additive models, ridge regression, variable selection, lasso
• High dimension, low sample size: principal components, partial least squares, regularization
• Supervised classification: logistic regression, neural networks, trees, forests, support vector machines, ensembles
• Unsupervised classification: hierarchical clustering, k-means, self-organizing maps, model-based clustering
• Cleaning, exploring and visualizing data

• Week 1. (Jul 27) Introduction to business analytics and R (Rob & Souhaib)

  • Lecture 1: What is Business Analytics?
  • Lab 1: R exercises
  • Lecture 2: Show case of R.

• Week 2. (Aug 3) Statistical Learning. Ch2. (Rob & Souhaib)

  • Lecture 3: More on R and statistical learning.
  • Lab 2:
  • Lecture 4: Assessing model accuracy. Bias-variance tradeoff

  Content:

  • The four V’s of big data/data science
  • Analytics and data science jobs: “By 2018, the US could face a shortage of up to 190.000 workers with analytical skills” McKinsey

• Week 3. (Aug 10) Regression for prediction. Ch3 (Rob) [Souhaib away]

  • Lecture 5: Review of linear regression, matrix formulation
  • Lab 3:
  • Lecture 6: Subset selection, LOOCV

• Week 4. (Aug 17) Resampling. Ch5 (Rob)

  • Lecture 7: Cross-validation
  • Lab 4:
  • Lecture 8: Bootstrap

• Week 5. (Aug 24) Dimension reduction. Ch6,10. (Rob and Souhaib)

  • Lecture 9: PCA
  • Lab 5:
  • Lecture 10: PLS

• Week 6. (Aug 31) Visualization. Own lecture notes. (Di)

  • Lecture 15: Basic data plots, categorical/numeric variables, facetting
  • Lab 8:
  • Lecture 16: Data cleaning

• Week 7. (Sep 7) Visualization. Own lecture notes. (Di)

  • Lecture 17: Plotting geographic data
  • Lab 9:
  • Lecture 18: Plotting multivariate data

• Week 8. (Sep 14) Classification. Ch4,8 (Souhaib & Di)

  • Lecture 11: LDA
  • Lab 6:
  • Lecture 12: Trees

• Week 9. (Sep 21) Classification. Ch4,9. (Souhaib) [Di away Sep 23-25]

  • Lecture 13: SVM
  • Lab 7:
  • Lecture 14: k-NN

• [Break Sep 28-Oct 4]

• Week 10. (Oct 5) Advanced classification. Ch8. (Di) [Rob and Souhaib away]

  • Lecture 19: Bagging, random forests
  • Lab 10:
  • Lecture 20: Boosting

• Week 11. (Oct 12) Advanced regression. Ch6. (Di & Souhaib) [Souhaib away?]

  • Lecture 21: Regularization
  • Lab 11:
  • Lecture 22: Shrinkage

• Week 12. (Oct 19) Clustering. Ch10. (Souhaib & Di) [Souhaib away?, Rob away Oct 22-]

  • Lecture 23: k-means
  • Lab 12:
  • Lecture 24: hierarchical clustering

James, Witten, Hastie, Tibshirani, An Introduction to Statistical Learning and Applications in R, Springer, 2013 (pdf available at http://www-bcf.usc.edu/~gareth/ISL/)

• Xie, Yihui Dynamic Documents with R and knitr Chapman and Hall/CRC 2013 http://yihui.name/knitr/

• Wickham, Hadley ggplot2: Elegant Graphics for Data Analysis Springer 2009 http://ggplot2.org/

• Cook, Swayne Interactive and Dynamic Graphics for Data Analysis: With Examples Using R and GGobi Springer 2007 http://www.ggobi.org/book/

• Chang, Winston R Graphics Cookbook O'Reilly 2013 http://www.cookbook-r.com/Graphs/

• Grolemund, Garret Hands-On Programming with R O'Reilly 2014 https://github.com/garrettgman

• Journal articles/web links: Analyzing the housing bubble in San Francisco http://vita.had.co.nz/papers/bay-area-blues.pdf

  ....

• R/RStudio (these packages with all the dependencies)

  • ggplot2, scales, stringr, plyr, dplyr, tidyr, reshape2, GGally, dichromat, magrittr, munsell, RColorBrewer, colorspace, wordcloud, vcd, gridExtra, hexbin, ggdendro, shiny, ggvis, ggsubplot
  • MASS, e1071, caret, randomForest, kohonen, cluster, fpc, mclust, rpart, nnet, nlme, vegan, penalizedLDA, PPTree, class, FNN, RSNNS, DMwR
  • mvtnorm, mvnormtest, HH, ICSNP, matrixStats, schoRsch, Matrix, psych
  • lubridate, tm, tuneR, caTools, maps, ggmap, maptools, shapefiles, sp, rworldmap
  • rmarkdown, knitr, devtools, roxygen2, profr
  • gwidgets, RGtk2, MissingDataGUI
  • foreign, jsonlite, curl, RCurl, rvest, Rcpp, XML, twitteR
  • xtable
  • classifly, clusterfly, meifly, LCA, LDAvis, nullabor
  • rggobi (requires ggobi software installed on the computer): installation should be easy on Windows machines

• Weka ? -- perhaps just mention that software other than R exists. It is too much to get them to learn more than one language in this unit.

Use Rmarkdown lecture notes like http://dicook.github.io/stat585/, and this should help students get comfortable with R themselves. (Ignore the ugly web site design!)

and material at this site: http://www.stat.iastate.edu/ccgs/short-courses/

Use RStudio, create a new project in your git directory (BusinessAnalyticsCourse), and open the slides/1-intro.Rmd.

On the interface you should see a "knitHTML" button. Click this to compile and preview the slides.