/ Recent content on Sean Conroy Hugo -- gohugo.io en-us Fri, 02 Apr 2021 00:00:00 +0000 /2021/04/02/r-w-data-table/ Fri, 02 Apr 2021 00:00:00 +0000 /2021/04/02/r-w-data-table/ Forgive this rant at the tidyverse universe. I started out learning R from Roger Peng’s excellent intro on Coursera. However, it wasn’t long before the tidyverse was being suggested as the ideal workflow for data science. Everything I could possibly want was offered by Hadley and Co. at RStudio. Need to plot stuff? There’s ggplot2. Need to work with dates? There’s lubridate. Need models? There’s parsnip (now caret). Need data manipulation? /2021/02/22/graph-feature-identification-with-1d-conv-nets/ Mon, 22 Feb 2021 00:00:00 +0000 /2021/02/22/graph-feature-identification-with-1d-conv-nets/ Motivation: Often in my line of work I’m asked to identify “kinks”, curves or bends in XY graph data, and I’ve spent inordinate amounts of time coding up various algorithms to find the change in the slope or threshold a change in the deviation around a linear fit, etc. These algorithms are never terrible accurate, and after taking most of Andrew Ng’s excellent Conv Nets class on Coursera last year, I’ve often wondered if CNN’s could be configured to be used for this type of work. /2021/01/23/crime-rates-in-dallas/ Sat, 23 Jan 2021 00:00:00 +0000 /2021/01/23/crime-rates-in-dallas/ To explore the Dallas Crime data more yourself, check out my Dallas Crime Explorer Shiny app here. The code behind this blog post is on GitHub here. First, let’s calculate the Crime Rates: PI[,Date := as.Date(substr(date1,1,10))] PI <- PI[Date >= as.Date("2014-06-01"),] setorder(PI,Date) PI[,MonthDate := as.Date(paste0(format(Date,"%Y-%m"),"-01"))] PI[grepl("MURDER",offincident) | grepl("HOMICIDE",offincident) | grepl("MURDER",ucr_offense) | grepl("HOMICIDE",nibrs_crime_category) | grepl("MURDER",nibrs_crime), NumMurdersPerMonth := .N,by = MonthDate] PI[,NumMurdersPerMonth := mean(NumMurdersPerMonth,na. /2021/01/14/shiny-apps-for-analyzing-dallas-police-data/ Thu, 14 Jan 2021 00:00:00 +0000 /2021/01/14/shiny-apps-for-analyzing-dallas-police-data/ Exploring Dallas Murders Forecasting Dallas Crime Rates Exploring Dallas Trafficking To explore the Dallas Crime data more yourself, check out my Dallas Crime Explorer Shiny app here. To explore the Dallas Murder data more yourself, check out my Dallas Murder Shiny app here. Dallas Trafficking To explore the Dallas Trafficking data yourself, check out my Dallas Trafficking Shiny app here. /2020/12/22/mapping-murders-in-dallas/ Tue, 22 Dec 2020 00:00:00 +0000 /2020/12/22/mapping-murders-in-dallas/ To explore the Dallas Murder data more yourself, check out my Shiny app here. The code behind this blog post is on GitHub here. Note, 1-11-2020: Data has been refreshed to include all of 2020. ## Warning: package 'gifski' was built under R version 4.0.3 All Dallas Murders since 2014, Color by Year Let’s take a look at the Dallas murders on a map for each year. Let’s take a look at the murders just for 2020. /about/ Wed, 02 Dec 2020 00:00:00 +0000 /about/ This is the personal website of Sean Conroy. My full resume is here. My interests are in applying statistics and machine learning to big hairy problems, and seeing if anything sticks. You can contact me at seantconroy at gmail.com /2020/12/02/murder-rates-in-dallas/ Wed, 02 Dec 2020 00:00:00 +0000 /2020/12/02/murder-rates-in-dallas/ Data Source Based on the following data set ‘Police Incidents’ from Dallas Open Data. To explore the Dallas Murder data more yourself, check out my Shiny app here. The code behind this blog post is on GitHub here. Data refresh date: 1-9-2021 Load data directly from DallasOpenData using Socrata API: PI <- read.socrata("https://www.dallasopendata.com/resource/qv6i-rri7.csv") setDT(PI) Analysis Ok, let’s get to work. First, let’s extract murder incidents by looking for “MURDER” or “HOMICIDE” in various columns used to describe the incident.