University of Florida

ABE 6933
Data Diagnostics:

Detecting and Characterizing Deterministic Structure in Time Series Data

Semester Taught - Fall

Description

Credits: 3

Application of nonlinear time series analysis to detect, characterize, and model deterministic structure in real-world time series data. Topics include signal processing, phase space reconstruction, surrogate data testing, causal network analysis, and phenomenological modeling.

In the process of data analysis, the investigator often observes highly volatile and random-appearing data. A common assumption is that observed volatility is due to underlying stochastic processes, but this is not necessarily the case. Nonlinear time series analysis (NLTS) allows researchers to test whether observed volatility conceals systematic nonlinear behavior, and to rigorously characterize governing deterministic dynamics. Behavioral patterns detected by NLTS, along with scientific principles and other expert information, guide the specification of mechanistic models that serve to ‘ground-truth’ and explain real-world behavior.

Pre-requisites

 

Course Objective

The tools of NLTS were developed in mathematics and physics. This course helps non-mathematicians in the applied sciences, engineering, economics, and other social sciences to become operational with NLTS.  Students acquire background knowledge of nonlinear dynamics required to apply NLTS in a sophisticated manner.  Students gain hands-on experience with NLTS so that they can apply it confidently to diagnose the dynamic forces driving volatile real-world data.

These objectives will be accomplished through:

  1. A ‘workshop’ classroom format emphasizing ‘knowledge through discovery’:  Students read assigned introductory material on scheduled topics before class. The instructor begins class by reviewing this material, punctuating it with intuitive examples and real-world applications, and answering questions.  Students then spend the majority of class time running prepared computer experiments under the instructor’s direct supervision to gain hands-on experience with NLTS.
  2. Detailed R code provided for computer experiments:  The R code used in computer experiments is explained in detail both by the instructor in the classroom, and by required readings.  This allows students to adjust the code for use in their own work.
  3. An explicit framework for applying NLTS methods to real-world time series data: The framework is condensed from sound empirical practices recommended in the literature.  Students become `data detectives', accumulating hard empirical evidence directing scientific inquiry.
  4. Homework projects that apply NLTS diagnostics to real-world time series data: Classroom computer experiments are supplemented with homework projects giving students increased hands-on experience with real-world data diagnostics.
  5. Evaluation of student skills with hybrid written and oral examinations:  Examinations test the extent to which students can apply NLTS methods to real-world data, and correctly interpret diagnostics results.  Each student provides a written report on data diagnostics, and further meets individually with the instructor to provide an oral defense of diagnostics and conclusions.

Instructor

Dr. Ray Huffaker
Office: 281 Rogers Hall
Phone: 392-1864 x281
Office Hours: Mondays 1-4pm or by appointment
rhuffaker@ufl.edu

Material/Supply Fees

None.

Class Materials Required

R. Huffaker, M. Bittelli, and R. Rosa. Nonlinear Time Series Analysis with R. Oxford University Press. The required textbook is coauthored by the instructor. It will be published by Oxford University Press (OUP) in October 2017 for international distribution. In the meantime, OUP has agreed to provide 20 pre-publication bound copies of the book at copying/binding cost to students in the course through the UF Follett Store. The instructor will receive no financial benefit from these sales. There are no additional fees for this course.

Course materials are available to students in Canvas.  Materials include lecture notes and slide presentations, the R code used in classroom computer experiments, homework assignments, and examinations.  Students are required to bring their personal laptops to class, download the most recent version of R, and dedicate time outside of class to familiarize themselves with programming basics in R

Homework Assignments and Examinations

There are three homework exercises assigned in Weeks 5,7, and 11, respectively, that are due two weeks after assignment. Students are encouraged to collaborate on homework assignments, but must turn in their own work. The course has a midterm examination composed of a written report and an oral defense. The written portion (assigned in Week 8) is take-home and open book, but students must work alone—collaboration is not permitted. Each student defends his/her written report in a fifteen-minute (private) oral defense in the instructor’s office (Week 10). The final examination also requires a written report (assigned in Week 13), and an oral defense (Week 16). The final follows the same procedure and guidelines as the midterm.

    

Course Outline

Topics (click for complete syllabus details)

Topics
Why Study Nonlinear Time Series Analysis?
Data Preprocessing
Data Preprocessing continued
Surrogate Data Testing
Empirically Detecting Causality
Empirically Detecting Causality with Heterogeneous Data
Phenomenological Modeling
Capstone Application of NLTS to Multi-Strain Infectious Diseases
Extreme Value Statistics

Grading

Assignment

Total Points

Percent of Grade

Homeworks (3)

100 (33.33 points/homework)

33%

Midterm (Written)

60

20%

Midterm (Oral)

40

13.33%

Final (Written)

60

20%

Final (Oral)

40

13.33%

Total

300

100%

Grading Policy:
A (269-300 points, 90-100%,); A- (254-268 points, 85-89%); B+ (239-253 points, 80-84%);
B (224-238 points, 75-79%); B- (209-223 points, 70-74%); C+ (194-208 points, 65-69%);
C (179-193 points, 60-64%); C- (164-178 points, 55-59%); D+ (149-163 points, 50-54%);
D (134-148 points, 45-49%); D- (119-133 points, 40-44%); E (0-118 points, 0-39%)

More information on UF grades and grading policies is located at: http://catalog.ufl.edu/ugrad/current/regulations/info/grades.aspx

Class Attendance and Make-Up Policy
Class attendance is essential for students to benefit from the classroom workshop approach.  Students should arrange with instructor for make-up material.  General UF policy can be found at:  https://catalog.ufl.edu/ugrad/current/regulations/info/attendance.aspx.

Academic Honesty

As a student at the University of Florida, you have committed yourself to uphold the Honor Code, which includes the following pledge: “We, the members of the University of Florida community, pledge to hold ourselves and our peers to the highest standards of honesty and integrity.”  You are expected to exhibit behavior consistent with this commitment to the UF academic community, and on all work submitted for credit at the University of Florida, the following pledge is either required or implied: "On my honor, I have neither given nor received unauthorized aid in doing this assignment." 
 
It is assumed that you will complete all work independently in each course unless the instructor provides explicit permission for you to collaborate on course tasks (e.g. assignments, papers, quizzes, exams). Furthermore, as part of your obligation to uphold the Honor Code, you should report any condition that facilitates academic misconduct to appropriate personnel. It is your individual responsibility to know and comply with all university policies and procedures regarding academic integrity and the Student Honor Code.  Violations of the Honor Code at the University of Florida will not be tolerated. Violations will be reported to the Dean of Students Office for consideration of disciplinary action. For more information regarding the Student Honor Code, please see: http://www.dso.ufl.edu/sccr/process/student-conduct-honor-code.  

UF Counseling Services

Resources are available on-campus for students having personal problems or lacking clear career and academic goals which interfere with their academic performance. These resources include:

  1. University Counseling Center, 301 Peabody Hall, 392-1575, personal and career counseling;
  2. Student Mental Health, Student Health Care Center, 392-1171, personal counseling;
  3. Center for Sexual Assault/Abuse Recovery and Education (CARE), Student Health Care Center, 392-1161, sexual assault counseling;
  4. Career Resource Center, Reitz Union, 392-1601, career development assistance and counseling.