普通人的数据科学

大数据分析近来已经成为了一个流行名词，明显很多公司已经在这个领域伸展拳脚了。其中预警的定位到底在哪里？预警对谁的好处最大？

“Big Data Analytics” is turning out to be quite the buzzword lately- and it is becoming apparent that there are many players in thisspace. So where exactly does Prelert fit? Who is Prelert reallybeneficial for?

First and foremost, the goal of Anomaly Detective^® isto bring sophisticated anomaly detection capabilities to themasses. We’re not a giant library of algorithms for datascientists. Heck, I'm not a data scientist (andmaybe you're not either) - I wouldn't know what do with 30different linear regression algorithms that one might find in a"Big Data Analytics" toolkit. But, if I have gobs of machine dataand I need some help finding behaviors that clue me in to why mysystems or users are misbehaving, I'd opt for something likeAnomaly Detective that will give me easy to use and focused anomalydetection.

Specifically, Anomaly Detective specializes in the detection of thefollowing four kinds of anomaly types:

 

• Unexpected changes in theoccurrence rate of things - perhaps youtypically receive about  30 orders per hour onWednesday afternoons. Wouldn’t you want to automatically know ifyou’re suddenly receiving much less than that?

• Unexpected changes in thevalues of things - nevermind trying toconstruct baselines or thresholds for literally thousands ofperformance metrics. Let automated anomaly detection do that dirtywork for you.

• Rarethings - wouldn’t you want to know if asystem that usually does not have an FTP process running on itsuddenly has one running and is sending data outbound?

• Population/PeerOutliers - “bad guys” are “bad” becausethey do things that are different than the good guys do. Populationor peer analysis bubbles up “who’s different” in your population ofusers or systems.

These four kinds of anomalies satisfy a vast majority of use casesin IT Operations, Application Performance Management, Security, andother disciplines. So, if you want a simple and effective way todetect anomalous behaviors in temporal data, look no further.Download, install, and find anomalies in your data in literallyless than 5 minutes.

原文链接：http://info.prelert.com/blog/data-science-for-the-rest-of-us

戴尔APM应用性能管理 DrXray

http://blog.sina.com.cn/xrayonapm