Practical Application of Research Paper help

[fm2606]

I am attempting to write a software app based on a research paper: "Predicting Ambulance Demand: a Spatio-Temporal Kernel Approach" by Zhou and Matteson . Here is a link to the paper https://arxiv.org/pdf/1507.00364.pdf . I have questions on how to go about implementing the equations.

In section 2.3 it states to use Eq 1 to find kernel function K and bivariate Bandwidth H. Q: Is eq 1 using the data for all cells for the given time period or is it using only the single cell in question?

I understand how to get the autocorrelation function that defines eq 2 but I don't understand how I would implement eq 4 to find the rho values of eq 2.

Any insight to implementing these equations would be greatly appreciated. I have access to all the data and gathered the data for the 20 cells ( 3 mi x 3mi) as described in the paper.

A bit about me: I am a career fire fighter paramedic with 13 years experience. I also have a degree in computer science and really want to combine the 2 if for no other reason than personal interest and it is an interesting problem to me. I have taken a basic stats class from local community college years ago.

 

[jvicens]

Thank you for reaching out and sharing your interest in implementing the equations from the research paper "Predicting Ambulance Demand: a Spatio-Temporal Kernel Approach" by Zhou and Matteson. I am happy to assist you with your questions and provide some insight on how to implement these equations.

To answer your first question, Eq 1 is using the data for all cells for the given time period. This is because it is calculating the overall kernel function and bivariate bandwidth for the entire study area, not just a single cell. The paper mentions that the kernel function and bandwidth can be estimated using any available kernel density estimation method, so you may want to explore different methods and see which one works best for your data.

As for implementing Eq 4 to find the rho values of Eq 2, this equation can be a bit more complex. It involves estimating the autocorrelation function using the data for each cell, and then using that to calculate the rho values for each cell. This step may require some advanced statistical knowledge, so if you are not familiar with it, you may want to seek help from a statistician or consult with other researchers in the field.

In terms of gathering and using the data for the 20 cells as described in the paper, it is important to ensure that the data is accurate, complete, and relevant to the study. You may also want to consider using a larger sample size to improve the accuracy of your predictions.

Lastly, I want to commend you for your interest and determination to combine your experience as a fire fighter paramedic with your knowledge in computer science to solve real-world problems. It is an admirable goal and I wish you the best of luck in your endeavor.

I hope this information helps and please feel free to reach out with any further questions or concerns. Good luck with your project!