Post on 12-Nov-2015
description
Motivation and pre-requisitesJeffrey LeekJohns Hopkins Bloomberg School of Public Health
About this courseThiscoursecoversthebasicideasbehindmachinelearning/prediction
Whatthiscoursedependson
Whatwouldbeuseful
Studydesigntrainingvs.testsets
Conceptualissuesoutofsampleerror,ROCcurves
Practicalimplementationthecaretpackage
TheDataScientist'sToolbox
RProgramming
Exploratoryanalysis
ReportingDataandReproducibleResearch
Regressionmodels
2/11
Who predicts?Localgovernments>pensionpayments
Google>whetheryouwillclickonanad
Amazon>whatmoviesyouwillwatch
Insurancecompanies>whatyourriskofdeathis
JohnsHopkins>whowillsucceedintheirprograms
3/11
Why predict? Glory!
http://www.zimbio.com/photos/Chris+Volinsky
4/11
Why predict? Riches!
http://www.heritagehealthprize.com/c/hhp
5/11
Why predict? For sport!
http://www.kaggle.com/
6/11
Why predict? To save lives!
http://www.oncotypedx.com/enUS/Home
7/11
A useful (if a bit advanced) book
Theelementsofstatisticallearning
8/11
A useful package
http://caret.rforge.rproject.org/
9/11
Machine learning (more advanced material)
https://www.coursera.org/course/ml
10/11
Even more resourcesListofmachinelearningresourcesonQuora
ListofmachinelearningresourcesfromScience
AdvancednotesfromMITopencourseware
AdvancednotesfromCMU
Kagglemachinelearningcompetitions
11/11