huberf/stat3105-proj2

Assassination Classroom Exercise

Installation

This project uses the following dependencies in R.

• tidyverse
• ggplot2
• jsonlite
• sp
• rgdal (likely optional)
• dlm

Running

The prediction logic is located in bomb_plan.R, but to execute, the main.R file provides utilities to easily run prediction given an array of start times and start locations.

The steps to execute are thus:

1. Open main.R, and at the top, modify the start_time, start_latitude and start_longitude vectors with your data. start_time values should be of the form, "2020-08-18T17:50:42.000Z" and start_latitude and start_longitude should be floating point values of standard latitude and longitude values.
2. After adding data as above, run Rscript main.R and it will print to console a prediction in the following form.

   Day Index: 1
   
   PREDICTION 1:
   Location (lat,long):
   UTM Location (lat,long):
   UTC Time (seconds since epoch):
   
   PREDICTION 2:
   Location (lat,long):
   UTM Location (lat,long):
   UTC Time (seconds since epoch):
   

3. Alternatively, you can programmatically access the prediction by importing the easy_predict function from bomb_plan.R and then passing it three vectors of the form described in step 1. It will return a list of the form:

   list(bomb1=list(utm_latitude=..., utm_longitude=...,
                   latitude=..., longitude=...,
                   time=...),
        bomb2=list(utm_latitude=..., utm_longitude=...,
                   latitude=..., longitude=...,
                   time=...))
   

Performance

On my own computer, the script takes approximately 6 seconds to produce a prediction.