INTRODUCTION TO MACHINE LEARNING

Recent years have witnessed an unprecedented increase in the availability of information on social, economic and health-related phenomena. Today researchers, professionals and policy makers have access to enormous databases, containing detailed information on individuals, companies and institutions and use of mobile devices. Machine Learning, a relatively new approach to data analytics, which lies at the intersection between statistics, computer science and artificial intelligence, has involved rapidly over recent years in response to our need to analysis these so-called Big Data sets. In contrast to the more tradition approach of data analysis, which focuses on prior assumptions relating to data structure and the derivation of analytical solutions, Machine Learning techniques rely instead on a model-free philosophy development of algorithms, computational procedures, and graphical inspection of the data in order to more accurately predict outcomes. The underlying approach taken is then to “let the data speak for itself”. Computationally infeasible until very recently, Machine Learning is itself a product of the latest advancements in both Information Technology and computing power.

This introductory course offers an intensive overview of the standard Machine Learning algorithms currently applied to social, economic and public health data, using a series of both official and community written Stata, Python and R commands. The primary objective being to illustrate how Machine Learning techniques can be applied to search for patterns in large databases, which can subsequently be used by researchers, professionals and policy makers alike to make both decisions and predictions. As a by-product, the course also serves to increase awareness as to Python and Stata’s “joint” capabilities to derive knowledge and value from large and often ‘noisy’ databases, through the use of both official and user written Machine Learning routines developed, which to date still remain relatively unknown to the majority of users.

COURSE STRUCTURE
The online format of our introductory Machine Learning course has been divided into two distinct modules, allowing researchers already familiar with the arguments discussed in the first module to choose to participant only in the second module.

The first module offers participants an overview of Python and Stata’s Machine Learning capabilities for data management, data quality analysis; exploratory data analysis, feature engineering and Principal Component Analysis. The second module instead focuses of the following popular Machine Learning methodologies: Supervised Learning, data management for Supervised Learning, Predictive Models, Logistic Regression, Stepwise Regression, Decision Trees, Neural Networks and Hyperparameter Optimization and Model Validation.

In common with TStat’s training philosophy, each individual session is composed of both a theoretical component (in which the techniques and underlying principles behind them are explained), and an extensive applied (hands-on) segment, during which participants have the opportunity to implement the techniques using real data under the watchful eye of the course tutor. Throughout the course, theoretical sessions are reinforced by case study examples, in which the course tutor discusses and highlights potential pitfalls and the advantages of individual techniques. The intuition behind the choice and implementation of a specific technique is of the utmost importance. In this manner, the course leader is able to bridge the “often difficult” gap between abstract theoretical methodologies, and the practical issues one encounters when dealing with real data.

COURSE OUTCOMES

At the end of the course, it is expected that participants:

have an understanding of the fundamental concepts and principles of Machine Learning;

are able to use Stata, R, and Python for data exploration, visualization, and preprocessing data in a Machine Learning context;

can independently implement the popular Machine Learning algorithms using Stata, R, and Python;

have attained an understanding of “real world” data issues through this hands-on experience; and

are able to implement solutions (with the help of the Stata and Python routine templates specifically developed for the course) to real world issues using Machine Learning techniques.

This course has been specifically developed for both professionals, researchers and Ph.D students working in advertising, business and management, biostatistics, economics, marketing, public health and social sciences, interested in applying the latest Machine Learning techniques in Stata and Python to big data.

Participants are required to have a good working knowledge of:

introduction to statistics;
descriptive and exploratory data analysis;
probability;
random variables;
probability distributions;
sampling distributions;
parametric estimation;
hypothesis testing;
simple and Multiple Regression;

Those needing to refresh these concepts are referred to:

Dekking, F. M., Kraaikamp, C., Lopuhaä, H. P., & Meester, L. E. (2005). A Modern Introduction to Probability and Statistics: Understanding why and how (Vol. 488). London: Springer;

Freedman, D. (2011) Statistics. Viva Books; 4th Edition (January 1, 2011);

McClave, James T. and Sincich, T. (2008). Statistics. Pearson;

Ross, S. M. (2020). Introduction to probability and statistics for engineers and scientists. Academic press;

Wasserman, L. (2004). All of statistics: a concise course in statistical inference (Vol. 26). New York: Springer.

Familiarity with a statistical software package such as SAS, Stata or SPSS is required;

some programming experience will also be distinct advantage.

NOTE: Those wishing to participate solely in Module 2, must be confident with the arguments covered in the first Module. Due to the intensive nature of this course, the course leader will unfortunately have insufficient time to go over the material already discussed in Module 1.

MODULE I | DATA MANAGEMENT FOR MACHINE LEARNING AND UNSUPERVISED LEARNING

SESSION I: DATA MANAGEMENT FOR MACHINE LEARNING AND FEATURE ENGINEERING

Data Management

Data matrix creation
Feature transformation (or engineering)

Data Quality Analysis

Identifying and handling missing data
Outliers

Exploratory Data Analysis

Subsetting data
Principles of Exploratory Data Analysis

Feature Engineering and Feature Selection

Feature Engineering
Feature Selection

Principal Component Analysis

Scree Plots
Biplots

SESSION II: UNSUPERVISED LEARNING

Unsupervised Learning

Introduction to unsupervised and supervised learning

Hierarchical Clustering

Choice of the approach
Dendrograms

K-Means Clustering

Algorithms

Clustering Validation

Clustering validation methodologies

MODULE II | SUPERVISED LEARNING

SESSION I: DATA MANAGEMENT FOR MACHINE LEARNING AND FEATURE ENGINEERING

Supervised Learning

Why supervised learning is so important

Data management for supervised learning

Feature Engineering
Feature Selection

Predictive Modelling

Training Set
Test Set

Linear Regression

SESSION II: UNSUPERVISED LEARNING

Logistic regression
Stepwise Regression

Alternatives to Stepwise Regression

Decision Trees
Neural Networks
Hyperparameter Optimization and Model Validation
Ensamble Learning

SUGGESTED READINGS

Microeconometrics Using Stata, Volume I: Cross-Sectional and Panel Regression Methods, A. Colin Cameron and Pravin K. Trivedi, Second Edition, Stata Press (2022).

Microeconometrics Using Stata, Volume II: Nonlinear Models and Causal Inference Methods, A. Colin Cameron and Pravin K. Trivedi, Second Edition, Stata Press (2022).

The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Hastie, T., Tibshirani, R., Friedman, J., Springer (2009).

An Introduction to Statistical Learning, Gareth, J., Witten, D., Hastie, T., Tibshirani, R., Springer (2013).

An Introduction to R -Notes on R: A Programming Environment for Data Analysis and Graphics Version 4.2.1 (2022-06-23.

Python 3.10.8 documentation.

The 2023 edition of this training course will be offered online. The course programme consists of 2 Modules – each divided into 2 sessions for a total of 8 hours of lessons per Module.

The first Module will be held on a part-time basis on the 8th-9th of June and the second on the 6th-7th of July both from 10 am to 2:30 pm Central European Summer Time (CEST).

MODULE I: CODE D-EF46-OL (2 online sessions)
Full-Time Students*: € 475.00
Ph.D. Students: € 605.00
Academic: € 700.00
Commercial: € 940.00

MODULE II: CODE D-EF47-OL (2 online sessions)
Full-Time Students*: € 475.00
Ph.D. Students: € 605.00
Academic: € 700.00
Commercial: € 940.00

*To be eligible for student prices, participants must provide proof of their full-time student status for the current academic year. Our standard policy is to provide all full-time students, be they Undergraduates or Masters students, access to student participation rates. Part-time master and doctoral students who are also currently employed will however, be allocated academic status.

Fees are subject to VAT (applied at the current Italian rate of 22%). Under current EU fiscal regulations, VAT will not however applied to companies, Institutions or Universities providing a valid tax registration number.

The number of participants is limited to 8. Places will be allocated on a first come, first serve basis. The course will only be confirmed when at least 5 people have enrolled.

Course fees cover: teaching materials (handouts, Stata do files and datasets to be used during the course) and a temporary licence of Stata valid for 30 days from the beginning of the course.

Individuals interested in attending these courses must return their completed registration forms by email (training@tstat.eu) to TStat by the 29th of May to register for Module I and the 28th of June 2023 to register for Module II.