User:Lwebster

From REU@MU
Jump to: navigation, search

Personal Info

Lindsay Webster is a double major in Mathematics and Theatre Arts in the Honors Program at Marquette University. She will graduate in 2019. She is the Production Manager of Marquette's Helfaer Theatre, where she also serves as a performer and occasionally a set designer.

She recently made a probabilistic mathematical model to predict the winner of Best Musical at the annual Tony Awards (and as of June 10th, it officially works!!!). So if you're wondering how Mathematics and Theatre Arts combine, that's it.


Research Topic

Preparing Gaussian Stochastic Processes for Coupling Landslide Hazards

How GaSP can be used to predict landslide behaviors.


Weekly Log

Week 1 (5/29/18-6/1/18)

Overall Goal: Clarify project topic, clear up any confusions on the previous reading, read a different take on Gaussian Stochastic Processes

  • Attended REU Orientation
  • Met Dr. Elaine Spiller and discussed previously assigned reading from the week before (Chapter 1-2,2 of Gaussian Processes for Machine Learning, C. E. Rasmussen & C. K. I. Williams)
  • Set up lab accounts and learned how to access the Wiki page
  • Learned about Library resources
  • Met with Dr. Elaine Spiller and was assigned new readings regarding Gaussian Stochastic Processes for the next week
  • Read about The Random Function (RF) Model (Jerome Sacks and William J. Welch)
  • Read about the Random Function Parameter Estimation, Prediction Uncertainty, and Diagnostics (Jerome Sacks and William J. Welch)
  • Read about Gaussian Stochastic Processes Designs (Jerome Sacks and William J. Welch)
  • Read about Uses of Random Function Models (Jerome Sacks and William J. Welch)
  • Read about Choosing a Random Function Model (Jerome Sacks and William J. Welch)


Week 2 (6/4/18-6/8/18)

Overall Goal: Explore Gaussian Matlab script in 1d and 2d, re-examine accomplishments and challenges and update weekly goals as needed, begin research paper

  • Reviewed all readings from last week
  • Began Responsible Conduct of Research Training - all 3 Modules completed
  • Watched educational videos on GaSP and took notes
  • Began research paper - created a journal-style format to write in as project develops
  • Received Gaussian Processes Matlab script from Dr. Elaine Spiller and wrote down any questions to be clarified
  • Attended REU Meeting in which proper presentation techniques were discussed
  • Downloaded "Assessment of Landslides Susceptibility," a Thesis by Gabriel Legorreta Paulin, to utilize the data and software later
  • Met with Dr. Elaine Spiller to clarify project topic and questions. Plans set in place for future work. Weekly goals were reevaluated.
  • Downloaded Matplotlib in Python to learn graphing in Python
  • Began learning graphing and matrices in Python


Week 3 (6/11/18-6/15/18)

Overall Goal: Translate Matlab script to Python, explore the effects of different values of theta, find inbuilt function to optimize the parameters of the Python Gaussian Model

  • Learned basic Matrix operations in Python - vectors, matrices, multiplying, dividing, dot products, inverses, transposes
  • Learned basic graphs in Python - line plot, bar graph, histogram, scatter plot, stack plot, pie chart
  • SKLearn was downloaded into Python to utilize the inbuilt GaSP functions
  • Utilized online tutorials to create a test prior and posterior GaSP model (Created by Jan Hendrik Metzen)
  • Altered test model to be prepared to receive and operate on mudslide data
  • Created a Python function to output the predicted y value when prompted with a specific x value, based on the GaSP model
  • Updated Paper Log and sources
  • Attend Responsible Conduct of Research face-to-face training
  • Downloaded LaTeX to prepare writing final research paper (may use Overleaf instead)
  • Observed the effects of a changing theta (the higher the theta, the smaller correlation between predicted lines)
  • Optimized the parameters of the Python Gaussian Model and created python functions to output those optimized values
  • Met with Dr. Elaine Spiller to discuss next steps
  • Began converting paper log and sources into Overleaf
  • Began reading new article, "Diagnostics for Gaussian Process Emulators"


Week 4 (6/18/18-6/22/18)

Overall Goal: Read Gaussian Process Diagnostics paper, read about linear regression (online course notes, online tutorials, my old course notes, and the first 3 chapters of the Landslide Thesis), prepare practice presentation

  • Ran into an issue with the Landslide software, so those activities have been put on hold
  • Finished reading and taking notes on "Diagnostics for Gaussian Process Emulators"
  • Read the first 3 chapters of "Assessment of Landslides Susceptibility"
  • Read "Logit/Probit" Lecture Notes
  • Read section 12.8 Logistic Regression from "An Introduction to Statistical Methods and Data Anylsis" by Ott and Longnecker and completed practice problems 12.43-12.45
  • Continued reading on logistic regression
  • Attended REU meeting to discuss project successes/challenges and proper presentation techniques
  • Met with Dr. Elaine Spiller to discuss readings, overall project goals, and next steps
  • Began creating practice presentation to be presented on 6/28


Week 5 (6/25/18-6/29/18)

Overall Goal: Explore new emulator Matlab script, work on converting it to Python, give practice presentation

  • Continued creating practice presentation - sent to Dr. Elaine Spiller for comments
  • Watched and took notes on logit/probit videos
  • Updated paper log in Overleaf to reflect Week 4 research
  • Revised halfway presentation and practiced presenting
  • Looked at Hyunjung Lee's Gaussian Process Emulator code
  • Gave Practice Presentations for everyone in the REU
  • Met with Dr. Elaine Spiller and Hyunjung Lee to discuss her code
  • Began developing a similar code in Python


Week 6 (7/2/18-7/6/18)

Overall Goal: Program composite Gaussian Process model in Python

  • Continued working on the composite Gaussian model in Python
  • Ran into a bug in the code - officially debugged
  • Finished translating the code into Python
  • Attended REU lunch and learned more about LaTeX
  • Ran into another bug in the code
  • Updated paper and sources in Overleaf
  • Continued fixing bug


Week 7 (7/9/18-7/13/18)

Overall Goal: Finish up composite Python script, use inbuilt Gaussian functions to create a 2D array Gaussian, begin project poster

  • Fixed bug from last week
  • Updated Weekly Goals
  • Updated final project paper
  • Started adding standard deviations to Python script
  • Used Inbuilt Gaussian Functions to optimize parameters in new script based off Ksenia's approach
  • Began project poster
  • Began trying to create a script that uses inbuilt Gaussian Functions and accepts a 2D array of inputs


Week 8 (7/16/18-7/20/18)

Overall Goal: Finish 2D array code, apply inbuilt 2D array to composite Gaussian, finish project poster

  • Finished 2D array code using inbuilt functions
  • Continued work on first draft of project poster to submit to Dr. Elaine Spiller for comments
  • Began applying inbuilt 2D array to composite Gaussian
  • Finished 2D Gaussian
  • Implemented a code to print the outputs for prompted inputs in the 2D Gaussian
  • Continued trying to get the inbuilt GaSP functions to work on the composite Gaussian
  • Continued work on poster - nearly done


Week 9 (7/23/18-7/27/18)

Overall Goal: Work on getting composite GaSP to work on inbuilt Python packages, create final presentation, complete a first draft of research paper

  • Finalized and submitted research poster
  • Continued work on composite GaSP in inbuilt packages
  • Began final presentation PowerPoint
  • Updated paper in Overleaf
  • Began writing the introduction and conclusion to research paper
  • Finished first draft of final presentation
  • Made significant progress on composite GaSP with inbuilt packages
  • Began combining 2D array with composite GaSP


Week 10 (7/30/18-8/3/18)

Overall Goal: Practice final presentation, finalize research paper, make last minute progress

  • Participated in Poster Session
  • Practiced final presentation
  • Continued work on finalizing code
  • Continued writing final paper
  • Gave final presentation
  • Finished and submitted final essay