Subscribe to DSC Newsletter

How would you detect and stop this type of crime?

Trailhead parking lots are notorious for crime activity - including from bears. These remote, un-patrolled locations, hidden in dense forests, attract all sorts of bad guys. One original crime is thieves stealing the park fees paid by hikers. These fees consist of $5 bills inserted by trail users (hikers) in an envelope, and dropped in a locked mailbox at the trailhead. These wooden boxes are easy to break and repair, for the purpose of stealing some of the $5 bills. A smart thieve would probably visit 10 trailhead parking lots with lots of hikers once a week, at night, and steal 50% of the envelopes (stealing 100% would attract police's attention). It could easily generate a $50,000 income a year, with little work, and low risk - especially since there are so many agencies involved in charging and collecting these fees. A really smart thieve would steal only $5,000 a year from each agency, but would target 10 different agencies associated with 10 different, popular hikes.

I'm not saying this crime is happening, but it is a possibility. Another type of crime would be perpetrated by official employees collecting these fees, and keeping a small percentage for themselves.

How would you detect this crime? Installing web cams might be too expensive in these remote locations with no Internet, no cell phone service. And how to prevent this crime from happening in the first place? And how to make sure every hiker pays the fee - usually $5 per day. Let's answer some of these questions.

How to make all hikers pay the fee?

Or increase payment rate by 30%. Last time I did one of these hikes, you had to either have a pass (to display in your car), or put a $5 bill in an envelope and fill a form (write down your licence plate number to prove that you paid). Like most tech guys, I had no pen to fill the form (I never carry a pen as I write notes on my cell phone), I had no dollar bills (I pay with credit cards), and I had no idea how/where to get a pass, and which of many different passes was required or OK for that specific hike.

So I came up with the following ideas for these agencies to collect the fees:

  • Allow hikers to pay after the hike, once back home, via the Internet. Most people are honest and would pay.
  • Levy a tax to subsidy trail maintenance
  • Have people drop a business card or an email address (rather than a $5 bill), and email them an invoice after the hike. 
  • Have a fence, with a door that opens when you insert money (for very popular locations)

How to prevent fees from being stolen by trailhead burglars or bad employees?

  • Don't accept cash payments (checks might be OK but are subject to other issues that would make it worse than cash)
  • Use taxes to pay for these services
  • Centralize the collection of money across agencies (and then hire a data scientist to analyze and reduce fraud)

What are you ideas? And how would you check if this crime is actually occurring (and where) and eating a decent amount of the revenue? You might even create your own job, by contacting all these agencies and offering data science services to increase revenue / reduce fraud. Each of these local agencies might be too small to hire a full-time data scientist, but they could split the cost and have one centralized data scientist work for all of them.

Views: 1258

Comment

You need to be a member of AnalyticBridge to add comments!

Join AnalyticBridge

Comment by Vincent Granville on July 12, 2013 at 4:55pm

@Martin: Great honeypot solution - I like it a lot!

Comment by Martin Magnus on July 12, 2013 at 4:52pm
Record the serial numbers of 20 $5 bills, put each bill in the same type of envelope used at each location and then deposit each envelope into each of the trailhead boxes. At the end of the week, collect all the money and find how many of the $5 bills come back. Your loss is directly estimatable from your loss on the $100. If $80 came back, you lost 20%.
Comment by Phillip Burger on July 11, 2013 at 11:37am

Can’t change it if we don’t measure it. I’d initiate the project by printing a serial number/unique identifier on the envelope and on the receipt retained by the hiker. This practice isn’t as common as it used to be because there is an extra cost to embossing id numbers on the paper.

Consider just two problems for now. Thieving at the trailheads and thieving by employees.

I’d be inclined to use two samples, one for each of the two problems. The measure is “lost envelopes.”

Design the first sample. Use a stratification sampling strategy. Take 10% of high-crime trailheads and 2% of low-crime trailheads. Put measures in place to lower thieving. Next year, collect data from the same trailheads. Test for significance in the difference.

Design the second sample for the thieving by employees. Use clusters. Cluster by jurisdiction of land management office. Take maybe 40% of trailheads known to produce the greatest number of $5 payments. Have research team tally envelopes for half of these.

Have respective land management office tally other half using their usual and customary practice. Test for significance in the difference. If after testing there is sufficient evidence to conclude that thieving by employees does occur, consider purposefully relying on the Hawthorne effect.

Comment by Vincent Granville on July 11, 2013 at 11:19am

A reader wrote:

The first and biggest obstacle I see (with the current situation) is there is not enough data to determine if there is a problem that needs to be addressed. 

If it's a box that collects the money... and the only way to know how much is in the box is for someone to remove the money and enter the information into a database, etc... then you will never know with complete certainty what 100% should be. 

EXAMPLE: Someone will enter $100 collected from Trail ABC on 7/11/13 into a database. But there is no way to know if that was originally $200 and someone (anyone - thief or employee) took $100. 

If you can't know what 100% is... then you'll never know if you have a problem. 

Here's my answer

I disagree with your point of view. If you collect payment data (e.g. weekly collected funds broken down by collector ID, day and hike #) you will then be able to see variations that can not be explained by chance alone, if there is more than a small amount of fraud. 

Comment by Greg Dolniak on July 10, 2013 at 7:16pm

Solution No. 1:

Pay by sending a SMS message.  At the same time you can make the hikers register themselves, so in case of emergency, police/fire department would have a general idea who is out there and where is going.  Something in the form:

"2 people from 12345 to 12346"

In this simple message you would register: telephone number (implicitly), time and date (implicitly), number of people in the party, where they start the hike (12345 would mean the trailhead), where they're going (12346 would mean trailhead destination).  For sending the message, $5.00 would be deducted from their account.

As for the benefits for the data scientists... well, need not to say ;-)

On Data Science Central

Follow Us

© 2018   AnalyticBridge.com is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service