{
  "id": "1996-april-riding-the-a-train",
  "slug": "1996-april-riding-the-a-train",
  "url": "https://dwillis.github.io/uplink/article/1996-april-riding-the-a-train.html",
  "year": 1996,
  "month": "April",
  "monthNum": 4,
  "headline": "Riding the A-train",
  "kicker": "CAR trips: First adventures",
  "author_name": "Jim Morris",
  "author_title": "",
  "affiliation": "CNN",
  "summary": "Jim Morris recounts his experience utilizing Federal Railroad Administration safety databases to identify hazardous railroad crossings.",
  "keywords": [
    "railroad safety",
    "FRA",
    "data cleaning"
  ],
  "topics": [
    "Investigative Methods",
    "Public Safety",
    "Data & Databases"
  ],
  "technologies": [
    "FTP",
    "Internet"
  ],
  "full_text": "Like a bolt of lightning, it struck. It was my first day at NICAR Bootcamp, and I had just had a revelation of Biblical proportions: I really could use this stuff. I was working on a railroad safety story, and I was anxious to use my new skills on my return to Atlanta. I got on the phone with the systems operator at the Federal Railroad Ad- ministration. I knew the FRA maintained sev- eral databases on train safety, and I wanted to know how I could access the information. I was expecting a bureaucratic nightmare, but what I found was computer nirvana. The FRA maintains a page on the Department of Transportation's website. The page includes ftp versions of all of its safety databases for 1990-95. First, I downloaded the file layout and code sheets. Then Idownloaded the databases. When I opened the first file, voila, nothing. Something was wrong with the layout. I began experimenting with the layout values and quickly discovered that things didn't add up. My record lengths were all wrong. The layout sheet listed fields that just didn't exist! After much fussing, I finally got it all worked out. Until I checked the codesheet. Ahh, the codesheet. Not even close to reality. I got back on the horn with the FRA and had them fax me a copy of the incident/accident report sheet from which the data is entered. Having the report form saved me from near insanity. By now, my eyes were burning, my fingers cramped. But I finally was ready to ask some questions. I wanted to know where the most deadly railroad crossing in the United States was. So I asked the computer to count the number of accidents at each highway crossing. That was a mistake. As I've since realized (I feel stupid admitting I didn't catch this right away), high- ways often cross railroad tracks more than once. Then I noticed the column marked \"GXID,\" or Grade Crossing ID. That was the magic field. Within seconds, I had the correct answer. IfI' ve learned anything from this, it is DON'T GIVE UP. Whenever I got stumped, I'd take a break, get some air. If I was still stumped, I'd call NICAR or the FRA and ask for help. I even believe this experience changed my life. As my wife will attest, I'm no longer afraid to ask for directions or admit I'm lost while driving my car. That's progress.",
  "provenance": {
    "extracted_by": "pre-v2",
    "extracted_at": "",
    "verified": true
  },
  "similar": [
    {
      "slug": "2000-may-railroad-safety",
      "score": 0.632
    },
    {
      "slug": "1996-june-tracking-rail-safety",
      "score": 0.496
    },
    {
      "slug": "2001-novemberdecember-tracking-rail-accidents",
      "score": 0.336
    },
    {
      "slug": "1999-januaryfebruary-hidden-hazards-618",
      "score": 0.204
    },
    {
      "slug": "2001-mayjune-truck-safety",
      "score": 0.111
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/1996-april-riding-the-a-train.html",
  "json_url": "https://dwillis.github.io/uplink/article/1996-april-riding-the-a-train.json"
}