United States professors create their own "criminal algorithm": prediction accuracy relies on data-crime, algorithm-IT information

Use algorithms to determine whether a person is enough science?

From United States Pennsylvania University of Richard Berk Professor spent for decades years of time since create has a on "crime behavior forecast" of algorithm, this sets algorithm can forecast out various crime related of content, for example which species people easy in adult Hou crime, which class criminals should was shut in prison of what regional, which species criminals again crime of chances high,, these algorithm of accuracy need based on big data.

In February 2013, the United States State of Wisconsin La Crosse police reportedly use this algorithm to sentencing the offender Eric Loomis, his car was used in a drive with shooting, the criminal was sentenced to six years in prison, suspended for five years. However, this decision has been criticized because of doubts about the scientific nature of the algorithm.

Specifically, the algorithm can be used widely, including judging which prisoners need to be locked in a restricted area, to how to choose a close monitoring of parole parole prisoner, police to predict which people arrested for domestic violence will again, for occupational safety and health administration provides about what information might be in violation of safety regulations in the workplace ... ... From birth to collect data to predict whether a person at the age of 18 may be crimes algorithm is currently being prepared, but for larger data requirements are very high. Not enough data to support will affect the accuracy of the algorithm.

As for the scientific nature of the algorithm, there cannot be verified, it may also need the support of big data. Before drawing substantive conclusions, one should not jump to comments.

美国教授自创“犯罪算法”：预测准确度依靠大数据 - 犯罪,算法 - IT资讯

使用算法来判断一个人是否够科学？

来自美国宾夕法尼亚大学的Richard Berk教授花费数十年的时间自创了一套关于“犯罪行为预测”的算法，这套算法可以预测出各种犯罪相关的内容，例如哪种人容易在成年后犯罪，哪类罪犯应该被关在监狱的何种区域，哪种罪犯再次犯罪的几率较高等，这些算法的准确度需要基于大数据。

在2013年2月，美国威斯康星州拉克罗斯警方被曝利用这一算法对惯犯Eric Loomis进行量刑，他驾驶的汽车曾在一起驾车枪击案件中使用过，这名罪犯最终被判处六年有期徒刑，缓刑五年执行。不过这一判决也受到了非议，原因就是外界质疑这一算法的科学性。

具体来说，该算法适用范围比较广泛，包括判断哪些囚犯需要关在限制区域，假释部门如何选择密切监视假释的囚犯，警方预测哪些因为家庭暴力而被捕的人会再犯，为职业安全与健康管理局提供关于哪些工作场所更可能会违反安全规定的信息……至于从出生时收集的数据预测一个人到18岁时是否可能会犯罪的算法，目前正在编写中，不过这对于大数据的要求非常高。没有足够的数据支持，必然会影响算法的准确度。

至于这种算法的科学性，目前还无法验证，这可能同样需要大数据的支持。在得出实质性的结论前，任何人也不应该妄下评论。