How to judge whether a statistic system exists leakage statistics By contrast method

Intermediary transaction http://www.aliyun.com/zixun/aggregation/6858.html ">seo diagnose Taobao guest cloud host technology Hall

In fact, in order to correct a wrong view: IP is missing statistics.

The correct view should be: PV Less is the leakage statistics.

PV is the page browsing, it is obvious, regardless of the behavior of these browsing behavior who is, as long as the page is browsed, statistical system should be counted to a PV.

So if a B two statistic system on the same site statistics, A is considered to be 10 PV, and B is considered to be 11 PV,

So it is obvious that a page has been overlooked.

So what is the difference between IP?

Now let's assume that the two statistic systems have the same PV (if there is a PV less, there must be a leak, that is not good contrast, and a statistics to 3 IP, and b statistics to 2 IP, at this time if you think is B leakage statistics of an IP behavior, then imagine, b missed this IP should at least one browsing? If B misses the behavior of this IP, then B should only count to 9 PV instead of 10.

Now that B is counting to 10 PV, it shows that B doesn't miss any of the pages being accessed.

So why is a statistic more than B statistics to an IP?

In fact, this can be understood, the same is 10 PV, A is considered to be 3 IP, in fact, the 10 PV in fact 7 times a "has visited the IP browsing behavior" and the same truth, B is that there are 8 times are "have visited the IP browsing behavior", that is to say, there is one browsing behavior, A thought is a new IP behavior, and b think is a once visited IP behavior, this is what the reason, it is obviously the time of memory B is relatively long.

Like what:

I want to remember all the IP behavior after the wee hours, as long as a certain IP after the early hours of the visit to your site, then this IP in the early hours of the next morning all browsing behavior is not considered a new IP generated behavior, because I want to think that this IP is not a new IP.

And some other statistical system memory time is not so long, an IP in the morning after visiting your site, to visit in the evening, the statistical system has not known this is a morning once visited the IP, so this browsing behavior is considered a new IP browsing behavior, Thus a more IP for this day.

How long does the recording of IP be used to differentiate whether it is a new IP, different statistical systems have different methods, so there will be different IP data.

I'll take it. The IP data is mainly used to answer the question "How many IP accesses are there today", so I'm going to think that a single IP should not be recorded as multiple IPs within a day. If the same IP in the morning visit and night access is recorded as two IP, then when you see the report on the 2 IP, you can not know whether this is an IP sooner or later two visits, or two IP access.

All statistical systems are likely to have leakage statistics, which is the result of a combination of several aspects, the need for statistical providers and webmaster to avoid the joint efforts.

Suspected that a statistical system exists leakage statistics of the mood can be understood, but using "IP is missing statistics," the "wrong way" to make a simple judgment is not sensible, it is easy to let oneself do not understand the real situation of IP volume.