Crime Prevention - Science topic

Jeus can give man eternal life and redemption, according to the Scritpure.

Yes - it seems to me that this case is much closer to typical events, but you will not get anywhere based on press information if you do not get comprehensive statistics of events of this type and analyze them in the context of the dynamics of crime changes in general. The fact that there was a road accident in my city today does not in any way indicate road safety.

Morning Banuka,

I ran a successful community cohesion programme ('The Grey Space Group'), while faced with being in command of what was reported in the media as the most violent police sub Division in Europe. Ultimately, we reduced serious violence by 40%.

I can provide links to relevant articles if of interest?

Kind Regards,

Martin

Tim and Martin

Interesting discussion, thank you. What kinds of rationality we use in different situational contexts is a relevant question.

Does it classify as "not rational" if someone is running from the police and kill someone else (by car) in the heat of the moment?

The interaction between individual characteristics (propensities) and environment incitements would need to test "some sort of rational behavior", but we would still depend on what we mean by rationality.

Situational Action Theory does not rely on the assumption that offenders are rational in the same sense as SCP does (the situation makes the criminal). They may be talking about two entirely different things as you both pointed out.

1.Accidents, Traffic - statistics and numerical data. 2.Accidents, Traffic - trends. 3.Wounds and injuries - epidemiology. 4.Safety - statistics and numerical data. 5.Data collection. I.World Health Organization. Dept. of Violence and Injury Prevention.

1 Social Impact Appraisal. 2 Accidents Impacts. 3 Physical Activity Impacts. 4 Security Impacts. 5 Severance Impacts. 6 Journey Quality Impacts and etc.

A Joint Report by the United Nations Office on Drugs and Crime and the Latin America and the Caribbean Region of the World Bank

UNITED NATIONS DEPARTMENT OF ECONOMIC AND SOCIAL AFFAIRS

Commission on Sustainable Development. Nineteenth Session. 2-13 May 2011

I provide some ideas and comments on data in this paper: https://rdcu.be/b48cvit

Philadelphia, PA

Dear Malik & readers,

You might want to go back over the discussion already at hand in the thread.

I think the answers are all there.

H.G. Callaway

---you wrote---

An interesting question waiting for the discussion to continue.

There are two answers I think:

1) k+2 where k is the number of predictors. This is the minimum required to derive an error term for the model. It won't generally be a very useful or precise model (but there may be exceptions)

2) it depends on how big the effects are you want to be able to detect relative to the error in the sample. This leads naturally to some kind of power/sample size estimate (or an analogous approach if you don't use significance tests)

Rules such as those suggested by Tabachnick & Fidell (actually Green's rule as they did not originate the formula) assume certain patterns of correlation among the predictors (probably independence) and a certain magnitude of effect relative to error. It is worth looking at Green's original paper to see if those assumptions are appropriate (I suspect generally not).

Green, S.B. How many subjects does it take to do a regression analysis? Multivariate Behavioral Research, 1991, 26, 499-510

I'm not sure it's directly related to assertiveness and self-esteem, but we wrote a paper a few year ago about the social dimension of medias and the importance to develop competences related to this dimension. Maybe you'll find some interesting ideas for your research.

Dear Nanda,

You can try Google's General Transit Feed Specification (GTFS) where you can find both static and real-time public transit data: https://developers.google.com/transit/

Best, Tao

A disclaimer first: I follow the field and have background in AI, but I have little hands-on experience with machine learning (ML), so some caveats apply.

I would say that detection of CAM might be borderline possible (with a significant error rate, but much better than random), but I would not advice on doing so. There are three main reasons:

1) Ethical implications: the same technology can be used to search for CAM by consumers, you are storing and manipulating a lot of sensitive data....

2) Current ML approaches to image/video classification are unreliable and would likely produce a large amount of false positives which may be a problem in itself - most content is not CAM, so even misclassifying 1 in million items wrongly will lead to huge amount of investigations into innocent people.

Worse, ML approaches can be tricked by an adversary (see e.g., https://openai.com/blog/adversarial-example-research/ and http://www.kdnuggets.com/2015/07/deep-learning-adversarial-examples-misconceptions.html) This could allow people distributing CAM to avoid detection by fooling the system to flag their content as 100% clean. An even worse problem is the potential use of materials that look OK to humans but are classified by the algorithm as CAM to put someone else in trouble. E.g, send someone modified cat pictures by email and then notify the police that the recipient has CAM. If the cat pictures were modified skillfully, the algorithm would confirm it is CAM, warranting further investigation of the recipient. "10 files on the computer of prime minister flagged as CAM" would make a sweet headline.

A nice example of similar (but simpler) effect in practice is unlicensed content on YouTube. YouTube has algorithms to discover when someone is posting copyrighted content. So the illegal content distributors started to modify the videos slightly (stretch, blur, letterbox, adjust audio...) and they get along mostly fine. And this is without using any of the sofisticated methods to fool the system that were mentioned above.

3) ML has bad reputation for repeating biases in the data. For example, if the training set CAM feature predominantly white/caucasian victims, the system may  easily learn that videos containing people with different skin tone are highly unlikely to contain abuse. Skin color is such a simple feature to learn that it is likely to be picked up if it correlates at least slightly with the target label. This could then result in systematic negligance in fighting CAM featuring victims from ethnic minorities.

More generally: implementing ML for such a task could be counterproductive, as it will reroute resources and focus from classification by humans. But the increased confidence of "we have an algorithm to find CAM" may easily prove unjustified as the criminals adapt to the way the system works.

If your question encompasses financial crime, then there are various techniques for detecting financial fraud using data mining.  Typically I use payment transactions data going through an organisation's accounts payable process (to suppliers) and payment transactions going through the payroll process (to employees).  I then set up various hypotheses about how fraud may have happened, usually in consultation with management using an action research approach (Kurt Lewin).  Mostly that requires adding further data sets such as GPS tracking data, cell phone data, building entry/exit data, CCTV etc.  Testing the data with those hypotheses often leads me to actual fraudulent transactions - sometimes different to what was expected under the initial hypotheses, but nearly always useful to management.  In the case of zero findings, that at least provides a degree of assurance that the fraud (probably) has not occurred during the period covered by the data.  The attached articles explain a little more fully.  Hope this helps.  

I wonder if those interested in Evidence-based Policing would have any insights.

Dear Adrein,

Given the times, the issue you are dealing with is not just topical but very imperative. My speciality is urban planning but just like everyone else violence, like an evil wind, touches everyone, one way or the other. I believe you have seen the piece on "Someone Stole My Smile: An Exploration into the Causes of Youth Violence in South Africa" edited by Patrick Burton (2007) < http://scholar.google.nl/scholar?q=faith-based+organization+and+curbing+violence+in+south+Africa+&btnG=&hl=en&as_sdt=0%2C5>.

Besides, a focus on the role of faith-based organisations is also key in violence reduction or prevention. For example, the Johns Hopkins' Bloomberg School of Public Health is currently doing a similar thing with its Faith-Based Violence Prevention Initiatives <http://www.jhsph.edu/research/centers-and-institutes/center-for-prevention-of-youth-violence/faith_based_initiatives>

I wish you all the best the best in your endeavour! 

This is one of the topics that are studied in 'environmental psychology'. They have done quite a lot of research on urban areas that feed crime. As far as I can remember, those areas with easy escape are more favorable to criminals. For example, in case of burglary, they prefer those houses which are close to parks or large open area. use the key word 'environmental psychology' in your searches and it will come up. Also you can search for it in two popular journals:1, Journal of environmental psychology and 2, Journal of Environment and behavior both have lots of articles regarding crime in urban and rural areas.

Dear Miranda,

To answer your question one needs to get statistical data which is not always easy. Regarding our situation we have been in recession for more than 15 years. Soft crime may be experienced in everyday life. Where I work at the university - an open public institution, anybody can enter it - one cannot leave open office doors because some steal money, computers even sun glasses etc. When travelling on train one should concern his/her personnel things.