Geographic Profiling, by cloudofwitnesses

[c4rtm4n]

Earlier this year I read about Geographic Profiling (GP) for the first time in a book about an unsolved case in Australia. I decided to learn a bit about GP so I could try and apply it myself. I learned quickly that this particular crime series was not big enough for an "accurate" geoprofile to be created. However, I found it a fascinating subject, particularly the work of Dr Kim Rossmo. I was looking for a distraction from real work and ended up developing a system to do geoprofiles utilising Rossmo's and other formulas. I came across EARONS at some point while looking for coordinate data and was aghast at the number of attacks. I've put together a website that shows various geoprofiles I have created, including the EAR crimes. The system is pretty crude compared to what I have read Rigel, CrimeStat etc are capable of. There is no crime weighting, buffer adjustment, ability to easily run scenarios etc, it just takes the coordinates, runs a formula and spits out a picture which I then overlay on google maps.

Here is a set of profiles with different formulas I did for the EAR attacks.

geographicprofiler.com/crimes/east-area-rapist-sacramento-area

Based on these profiles Rancho Cordova appears to be the hottest area, it has a high number of very nearby attacks.

However:

I've run scenarios of the first 10 attacks: hottest spot Rancho Cordova, avg attack location east of Carmichael.

The next group of attacks before the 3 month break, the hot spots are near Olsen Villa and north of Lake Knoll estates. The avg coords of attacks has moved closer to Carmichael.

The last group of attacks after the 3 month break has hot spots around Pasadena Estates that goes north east up Auburn Blvd to a place Google maps call Hemlock Condos. The average of the attack locations is now a couple of miles west of Carmichael about a mile north of Arden Arcade..

What does this tell us? TBH I don't know. I have wondered if it is the work of 1 perp or multiple perps/series..

So, please take this with a grain of salt, while I've researched EARONS to a degree, I really don't know much about the series. The coordinates I have used look about right from what I've read, but I really have no idea how accurate they actually are. Garbage in = garbage out.

Watch this interview from a Zodiac documentary with Dr Kim Rossmo www.youtube.com/watch?v=fXbCR0nGj_E to get a good quick summary of what GP is and what it is meant to achieve.

geographicprofiler.com/papers has links to the papers and articles that I found helpful. If your're interested in the formulas and GP in general these papers are an good introduction.

Anyway, I hope you find it interesting.

Cheers

[Any of N]

Nov 11, 2016 22:39:07 GMT c4rtm4n said:

Earlier this year I read about Geographic Profiling (GP) for the first time in a book about an unsolved case in Australia. I decided to learn a bit about GP so I could try and apply it myself. I learned quickly that this particular crime series was not big enough for an "accurate" geoprofile to be created. However, I found it a fascinating subject, particularly the work of Dr Kim Rossmo. I was looking for a distraction from real work and ended up developing a system to do geoprofiles utilising Rossmo's and other formulas. I came across EARONS at some point while looking for coordinate data and was aghast at the number of attacks. I've put together a website that shows various geoprofiles I have created, including the EAR crimes. The system is pretty crude compared to what I have read Rigel, CrimeStat etc are capable of. There is no crime weighting, buffer adjustment, ability to easily run scenarios etc, it just takes the coordinates, runs a formula and spits out a picture which I then overlay on google maps.

Here is a set of profiles with different formulas I did for the EAR attacks.

geographicprofiler.com/crimes/east-area-rapist-sacramento-area

Based on these profiles Rancho Cordova appears to be the hottest area, it has a high number of very nearby attacks.

However:

I've run scenarios of the first 10 attacks: hottest spot Rancho Cordova, avg attack location east of Carmichael.

The next group of attacks before the 3 month break, the hot spots are near Olsen Villa and north of Lake Knoll estates. The avg coords of attacks has moved closer to Carmichael.

The last group of attacks after the 3 month break has hot spots around Pasadena Estates that goes north east up Auburn Blvd to a place Google maps call Hemlock Condos. The average of the attack locations is now a couple of miles west of Carmichael about a mile north of Arden Arcade..

What does this tell us? TBH I don't know. I have wondered if it is the work of 1 perp or multiple perps/series..

So, please take this with a grain of salt, while I've researched EARONS to a degree, I really don't know much about the series. The coordinates I have used look about right from what I've read, but I really have no idea how accurate they actually are. Garbage in = garbage out.

Watch this interview from a Zodiac documentary with Dr Kim Rossmo www.youtube.com/watch?v=fXbCR0nGj_E to get a good quick summary of what GP is and what it is meant to achieve.

geographicprofiler.com/papers has links to the papers and articles that I found helpful. If your're interested in the formulas and GP in general these papers are an good introduction.

Anyway, I hope you find it interesting.

Cheers

Very nice work, cartman.  It's quite nice to be able to compare the results of multiple approaches. Thank you for sharing this with us.  I see this as "work in progress" and how things tend to look messy until a final result is selected and presented.  That's not intended as a knock on your work at all.  That's just how it works.

Supposedly, Rossmo's profile -- the one that he made himself-- pointed to " the area of Manzanita Avenue/Coyle Avenue between Madison Avenue and Auburn Boulevard."  That's roughly between points #5 and #18 on the FBI map (Victims 5 and 17).  It's hard for me to tell, but it looks like one of your hot spots for the Rossmo calculations jibes with that.

Looks like you have different numbers of clusters for different methods.  Was that your choice, or did you leave it up to the algorithms to decide? 

[c4rtm4n]

Thanks Any of N, no doubt about it, it is a work in progress. And I am not a Geographic Profiler. In terms of the different clusters, that is just the different formulas, I'm not doing any tweaking. If you look at some other cases I've processed geographicprofiler.com/crimes you can sort of get a feel for how the different formulas will work.

Rossmo's formula is all about the size of the buffer zone which is a configurable variable in his formula. From what I gather, the general rule for these type of crimes is to set the buffer to be half the average of the nearest neighbour distance - en.wikipedia.org/wiki/Nearest_neighbour_algorithm, however this will change depending on what subset of crimes are used in the calculation. If you zoom in on the bottom left corner of the geoprofile you'll see the buffer size printed in meters.

Something I found very interesting about the EAR attacks, is that the centrography profile geographicprofiler.com/crimes/east-area-rapist-sacramento-area/centrography at a glance looks close to a perfect circle. It's certainly closer to a circle than the other crime series I've processed. With a center around Carmichael, I do wonder why there were no attacks there.

[bob]

"In Davis, a police dog tracked the offender to a private airport. "

I hadn't heard that before. It may have been that he parked there as a means of deflection, but since this was Davis, would he have expected a police dog following his scent?

Private pilots have to stay current by logging some hours every so often. Do these attacks coincide with the time window used to meet that requirement?

Nice to see something new being explored.

[bobthebuilder]

Nov 14, 2016 4:27:07 GMT c4rtm4n said:

Thanks Any of N, no doubt about it, it is a work in progress. And I am not a Geographic Profiler. In terms of the different clusters, that is just the different formulas, I'm not doing any tweaking. If you look at some other cases I've processed geographicprofiler.com/crimes you can sort of get a feel for how the different formulas will work.

Rossmo's formula is all about the size of the buffer zone which is a configurable variable in his formula. From what I gather, the general rule for these type of crimes is to set the buffer to be half the average of the nearest neighbour distance - en.wikipedia.org/wiki/Nearest_neighbour_algorithm, however this will change depending on what subset of crimes are used in the calculation. If you zoom in on the bottom left corner of the geoprofile you'll see the buffer size printed in meters.

Something I found very interesting about the EAR attacks, is that the centrography profile geographicprofiler.com/crimes/east-area-rapist-sacramento-area/centrography at a glance looks close to a perfect circle. It's certainly closer to a circle than the other crime series I've processed. With a center around Carmichael, I do wonder why there were no attacks there.

Great work Cartman, I agree on centrography analysis, its almost remarkable how perfect the circle is around the Carmichael or RC. I have always felt he was in Carmichael based on my own (amateur) geo-analysis. Surely LE could use this as a good starting point and database any male in that area, then filter.

I hope you don't mind but I have added your analysis to wikipedia in the hope it may jog someones mind from a wider readership. Please let me know if you object I will remove it.

[Deleted]

Nov 18, 2016 1:59:06 GMT bobthebuilder said:

Nov 14, 2016 4:27:07 GMT c4rtm4n said:

Thanks Any of N, no doubt about it, it is a work in progress. And I am not a Geographic Profiler. In terms of the different clusters, that is just the different formulas, I'm not doing any tweaking. If you look at some other cases I've processed geographicprofiler.com/crimes you can sort of get a feel for how the different formulas will work.

Rossmo's formula is all about the size of the buffer zone which is a configurable variable in his formula. From what I gather, the general rule for these type of crimes is to set the buffer to be half the average of the nearest neighbour distance - en.wikipedia.org/wiki/Nearest_neighbour_algorithm, however this will change depending on what subset of crimes are used in the calculation. If you zoom in on the bottom left corner of the geoprofile you'll see the buffer size printed in meters.

Something I found very interesting about the EAR attacks, is that the centrography profile geographicprofiler.com/crimes/east-area-rapist-sacramento-area/centrography at a glance looks close to a perfect circle. It's certainly closer to a circle than the other crime series I've processed. With a center around Carmichael, I do wonder why there were no attacks there.

Great work Cartman, I agree on centrography analysis, its almost remarkable how perfect the circle is around the Carmichael or RC. I have always felt he was in Carmichael based on my own (amateur) geo-analysis. Surely LE could use this as a good starting point and database any male in that area, then filter.

I hope you don't mind but I have added your analysis to wikipedia in the hope it may jog someones mind from a wider readership. Please let me know if you object I will remove it.

What I find fascinating about the centrography report map showings , it is in the same block as the Patty Hearst (SLA)Crocker bank robbery and murder . Probably no connection,but you never know !

[morf13]

Is there a way to get a full list of the streets included within the most center part of the map??

[trabuco]

I also think this is great work. Now who will be looked for?

[bobthebuilder]

Interesting to see the power/control killer like to extend the period of the attack, interesting ERONS liked to do the same. The variation in jobs is really interesting, smacks of BTK and the suffolk strangler as highlighted.   

Video on profiling
www.youtube.com/watch?v=Fx5WHZoqGEI

[batman]

Is it the case that EARONS didn't pick places that are central because he preferred to escape into rural areas?

That this de facto makes his choices on the parameter of a population and therefore as he works around these parameters, a circumference without inner activity will form, even though he may have nothing to do with anything within the circle?

[suburbsirvine2]

Oct 15, 2016 6:45:39 GMT @cloudofwitnesses said:

We're on the same page. The UCDavis airport is just 1 mile from the 6/24 attack. I wonder if flight logs were kept/archived for either Modesto or Davis.

Flight logs for those small airports would be in a pilot's own log book, not public information.

[jimh]

May 27, 2017 9:24:22 GMT Tom Box said:

From my research the logs of the pilots that "flew" the areas that The Spink Corporation was mapping at that time might be worth looking at.

From your research lol. Well, since you researched, please disclose. We're waiting...

[spodey]

Nov 11, 2016 22:39:07 GMT c4rtm4n said:

Earlier this year I read about Geographic Profiling (GP) for the first time in a book about an unsolved case in Australia. I decided to learn a bit about GP so I could try and apply it myself. I learned quickly that this particular crime series was not big enough for an "accurate" geoprofile to be created. However, I found it a fascinating subject, particularly the work of Dr Kim Rossmo. I was looking for a distraction from real work and ended up developing a system to do geoprofiles utilising Rossmo's and other formulas. I came across EARONS at some point while looking for coordinate data and was aghast at the number of attacks. I've put together a website that shows various geoprofiles I have created, including the EAR crimes. The system is pretty crude compared to what I have read Rigel, CrimeStat etc are capable of. There is no crime weighting, buffer adjustment, ability to easily run scenarios etc, it just takes the coordinates, runs a formula and spits out a picture which I then overlay on google maps.

Here is a set of profiles with different formulas I did for the EAR attacks.

geographicprofiler.com/crimes/east-area-rapist-sacramento-area

Based on these profiles Rancho Cordova appears to be the hottest area, it has a high number of very nearby attacks.

However:

I've run scenarios of the first 10 attacks: hottest spot Rancho Cordova, avg attack location east of Carmichael.

The next group of attacks before the 3 month break, the hot spots are near Olsen Villa and north of Lake Knoll estates. The avg coords of attacks has moved closer to Carmichael.

The last group of attacks after the 3 month break has hot spots around Pasadena Estates that goes north east up Auburn Blvd to a place Google maps call Hemlock Condos. The average of the attack locations is now a couple of miles west of Carmichael about a mile north of Arden Arcade..

What does this tell us? TBH I don't know. I have wondered if it is the work of 1 perp or multiple perps/series..

So, please take this with a grain of salt, while I've researched EARONS to a degree, I really don't know much about the series. The coordinates I have used look about right from what I've read, but I really have no idea how accurate they actually are. Garbage in = garbage out.

Watch this interview from a Zodiac documentary with Dr Kim Rossmo www.youtube.com/watch?v=fXbCR0nGj_E to get a good quick summary of what GP is and what it is meant to achieve.

geographicprofiler.com/papers has links to the papers and articles that I found helpful. If your're interested in the formulas and GP in general these papers are an good introduction.

Anyway, I hope you find it interesting.

Cheers

Very interesting...Here is a link to the development of "geographic profiling" : en.wikipedia.org/wiki/Geographic_profiling

[morf13]

Rossmo did a good profile of Zodiac

www.youtube.com/watch?v=fXbCR0nGj_E

[whereishe]

Sept 12, 2016 18:35:25 GMT @cloudofwitnesses said:

The idea behind the broader estimate would be to remove the American River as an obstacle to factor in. I thought of this only as a last solution, but then it occurred to me that in 1976 and 1977 California was in a major drought, and the water level of the American River was very low. In fact, according to California Dept of Water Resources, the American River flow was at 31% of normal in 1976, and 14% in 1977. Here's a photo of the American River at that time:

s6.postimg.org/5mits26wx/American_River_70s_drought.png

This raises the possibility that the EAR could traverse the river without difficulty.

The American River seems to always be discussed as an obstacle. I'm not sure it was. Not for an avid bicyclist. Rancho Cordova connects to Fair Oaks via the American River bicycle path, just west of the Sunrise bridge. It was a viable path back in the 1970's. Fair Oaks adjoins Carmichael. Just a thought...