数据异常如何分析How to analyze data anomalies

gaowenhua

为什么涨和跌都要关注？

在数据分析的过程中，我们不仅仅要关心跌，以便采取相应动作，减缓跌的趋势，

也更要关心涨，弄清楚涨的原因，并放大它，或者说是复制它

gaowenhua

Why should we pay attention to both rising and falling?

In the process of data analysis, we not only need to care about the decline, but also take corresponding actions to slow down the downward trend,



We also need to pay more attention to the rise, understand the reasons for the rise, and amplify it, or rather replicate it

gaowenhua

发现异常

就像你发现昨天数据跟往前不一样，猛涨了还是猛跌了，通过观测数据发现异常

gaowenhua

Discovering anomalies



Just like how you noticed yesterday that the data was different from before, whether it surged or plummeted, by observing the data and discovering anomalies

gaowenhua

确定问题

发现异常之后，我们要确定这个异常是不是一个问题，有多严重，

从时间维度上进行周同比、月同比或者是年同比。

gaowenhua

Identify the problem



After discovering the anomaly, we need to determine whether it is a problem and how severe it is, and conduct weekly, monthly, or yearly comparisons from a time dimension.

gaowenhua

确定原因


对于这个异常的指标从不同的维度去拆解，找出原因。

gaowenhua

Determine the cause




Decompose this abnormal indicator from different dimensions and identify the cause.

gaowenhua

解决问题

找到原因之后，就是针对性的解决问题了，

根据问题的原因，去解决这个问题

gaowenhua

Solving problems



After identifying the cause, it is time to address the problem in a targeted manner,



Based on the cause of the problem, solve it