This document outlines the topics and approach for a statistics and data mining methods course. The course will cover 4 main topics over 4 lectures: exploratory data analysis, linear and logistic regression, classification trees and ensembles. Class participation and computer work will be graded, with originality and creativity rewarded. The seminars are intended to provide a short overview to spur further self-directed learning, as it is not possible to cover all material in depth. Students are encouraged to review concepts, think critically about problems, and consider alternatives.

1. Statistics and Data Mining
Methods.
Leonardo Auslender
Leonardo ‘dot’ Auslender ‘at’ gmail
‘dot’ com
732 494 1555.
September 2019

2. 4 Lectures:
1: EDA Principal Components Clustering
2: Linear Regression and Variable Selection
3: Logistic Regression and Variable Selection
4: Classification Trees, Gradient Boosting and ensembles.
5: Time permitting: Issues.
Grading policy: Class participation, discussion and computer work,
possibly homework. Originality and creativity are heavily rewarded, as long
as they are disciplined.
Since it is not possible to review all analytical aspects, emphasis is in
thinking on the problem, finding out satisfactory solution and in being
able to contrast it to alternatives.
Also not possible to focus in depth on material. Read at your leisure,
keep on reading, and focus on INTERVIEW questions. Some of the
questions require lots of research.

3. “If there are no alternatives, then we are failing to think about the
problem. Or else, there is no problem to begin with.”.
Seminars offer extremely short and terse view of many analytical
aspects of data analysis and problem conceptualization. Just a guide
of areas that need extensive further review.
Review, re-read, make mistakes, re-learn, etc.
TURN OFF YOUR CELL PHONES FOR THE
DURATION OF THE CLASS.