Training courses for PhD/Master students
Short Bio: Qidi Peng received his PhD degree of Applied Mathematics – Probability and Statistics at University of Lille. He then became a research assistant professor at Claremont Graduate University. His research field covers but is not limited to machine learning on stochastic processes and its applications to finance. Since 2016, he was hired as a data expert in American International Group. Since 2021 he became research associate professor at Claremont Graduate University.
Title: Statistical Learning: classification and prediction, with application in R
Abstract: In this mini lecture some most popular and cutting-edge machine learning approaches are introduced. These methods are motivated through solving a real world problem: modelling the behavior of out-of-pocket costs of healthcare insurance in the USA. The material involves introduction to some base learners of classification such as logistic regression, regularized logistic regression (ridge and lasso) and random forests. Data preprocessing methods are also introduced to perform data clean and variable selection. A voting ensemble learning method is designed to obtain the subset factors which can best explain the behavior of out-of-pocket costs. Note that similar analysis can be straightforward applied to totally different set of problems with similar data type. Implementation of R scripts is also presented and we will explain how to analyze the output.
Mehdi Aharchaou, Houston, TX, USA.
Short Bio: Mehdi Aharchaou has worked since 2013 in the Oil & Gas industry as a research scientist and technical lead at ExxonMobil. In his most recent role he managed R&D projects aiming to realize the full potential of Machine Learning and Data Analytics in exploration geoscience, to achieve goals such as cost reduction, streamlining processes and empowering decision making. Prior to that, as a senior research scientist he led the corporation's efforts in the design, development and deployment of several advanced data analysis and processing capabilities that involve signal processing, time series analysis, sparse and stochastic modeling etc. Mehdi published 20+ technical articles in peer-reviewed geoscience journals (GEOPHYSICS, TLE, GJI), submitted 5 patents, presented at top-tier geoscience conferences (SEG, EAGE, GSH), and he organized several workshops and trainings for the geoscience community. In 2020, Mehdi was nominated for the Clarence Karcher award, the most prestigious award a young geoscientist can receive in the field of applied geophysics. Mehdi holds three Masters degrees from Georgia Institute of Technology (Electrical & Computer Engineering; Atlanta, USA), Rice University (Geoscience; Atlanta, USA), and INP-ENSEEIHT (Electrical Engineering; Toulouse, France) with majors in Statistical Data Analysis & Processing. His research interests include Business Analytics, Data Science, Supervised / Unsupervised ML, AI, Numerical Modeling and Big Data.
Title: Advanced Analytics and AI Applied to Industry, Use Cases From Oil & Gas Sector
Abstract: The presence of artificial intelligence (AI) is impacting almost every industry. Unlike general AI which is a frontier research discipline to build intelligent agents that perform any task with human-level intelligence, industrial AI is more concerned with the practical application of such technologies to address industrial pain-points for customer value creation, predictive analytics, cost and cycle-time reduction, productivity enhancement and insight discovery. Large improvements in computing technology and massive increases in data volumes have allowed companies to build new data science and machine learning (ML) technologies to solve previously unsolvable problems and derive business value from data.
The aim of this course is to give an accessible overview of AI and explain how it relates to Data Science, Machine Learning and Data Analytics. Topics include: (1) Main stages of an end-to-end AI project, (2) Regression vs Classification; (3) Supervised learning (neural networks, deep learning, decision trees, random forests, support vector machines, ensemble methods), (4) Unsupervised learning (clustering, dimensionality reduction, recommender systems); (5) Challenges and best practices in machine learning (bias/variance tradeoff, cross validation, regularization, overfitting / underfitting). The course will draw from numerous case studies and applications of AI and ML in the Oil & Gas sector.