SGAS

This project was part of an REU undergraduate research summer program funded by NSF at Texas State University.

Disclaimer: This material is based upon work supported by the National Science Foundation under REU grant #1757893. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

File Usage

Graph5.m

This file is used to graph our results. You can set the simulation function to either SGAS5 of ML for either the Constant Density Approximation solution or the Machine LEarning solution respectively. To graph the Perfect Prediction, go to the SGAS5 file and un comment the temp = line and k = line to set k equal to the ground truth value.

realScipt.m

This file is used to create the labeled data to feed into the machine learning model for training. However we did the standardization of the inputs in Excel so that is not part of the script.

Octave: A Side Note

Most of the Matlab code can be ran in octave except for the reallife script which I made an Octave counter part script reallife_octave to be used if running the scripts in an Octave environment. The purpose of reallife is to take a CSV file of gas station data and convert it into a highway for the MatLab simulations to be run on.

scrape.py

This file converts html files pulled from GasBuddy.com Trip Planner into CSV files that our other scripts can read. eg. python scrape.py Trip*.html to scape all pages. However it will skip any files that already have a cooresponding CSV.

append.py

Adds an html comment at the head of a GasBuddy file to specify the location for easy access. eg. python append.py Trip*.py Will skip html files with comment as first line.

analyze.py

Prints out statistics for Trips. eg. python analyze.py Trip*.csv Must be run before label.py

label.py

First run analyze.py to get lengths.csv. Then simply run python label.py to label how many gas stations are in certain sections of the highway for all the analyzed trips.

plot.py

Plots one (1) trip with the x-axis as the position along the highway and the y-axis being the price of gas. This helps me visualize what a Trip looks like. Usage: python plot Trip32.csv

hist.py

Plots a histogram of gas prices for a Trip. This helps me see of the resulting gas price of a simulation run is a good price or not. Usage python hist.py Trip20.csv

machinelearning.py

Trains a machine learning model. Usage: python machinelearning.py inputs.csv inputs.csv is a CSV where each line is a data set. A data set consists of n inputs (prefereably standardized) and the last comma separated value being the label for the data. The machine learning script will then output a vector A (1xn) and a scalar value b. Now any input vector x (1xn) can be calculated into an output with the formula A * x' + b where x' is the transverse of x.

Printing out figures

The current general procedure for printing out the figures in the paper are as follows

1. Set style ':rs'
2. Run Graph5 with SGAS5 with temp and k commentd out
3. Set style ':gx'
4. Run Graph5 with SGAS5 with temp and k not commented out
5. Set style ':bo'
6. Run Graph5 with ML
7. run subplot(2, 3, 1)
8. run x = legend("Perfect Prediction", "Const. Density Approx.", "Machine Learning")
9. run set(x, "color", "none") for transparency purposes
10. run export_fig fig.png -transparent
11. You may then need to crop the figure because the bottom graphs are for changing the start of the critically low section

Note: Step 10 requires a package altmany-export_fig which you can simply download then place the folder in your MatLab directory then run the command addpath('altmany-export_fig-b1a7288') so that MatLab can find the required files.