发明名称 Statistical model for estimating unique users from unauthenticated cookies
摘要 This disclosure generally relates to systems and methods that facilitate employing a statistical model over a specified time frame divided into a plurality of time intervals for estimating a quantity of unique users from a set of unauthenticated unique identifiers, such as cookies, associated with accesses to one or more servers.
申请公布号 US9245228(B1) 申请公布日期 2016.01.26
申请号 US201313974838 申请日期 2013.08.23
申请人 GOOGLE INC. 发明人 Pihur Vasyl;Dijamco Armand;Diez David;Dirks William
分类号 G06N5/02;G06N7/00 主分类号 G06N5/02
代理机构 Lowenstein Sandler LLP 代理人 Lowenstein Sandler LLP
主权项 1. A method, comprising: accessing, by a device including a processor, a plurality of unauthenticated unique identification records associated with transactions between at least one client device and at least one server device during a specified time frame, wherein respective unauthenticated unique identification records are associated with respective unauthenticated unique identifiers of a plurality of unauthenticated unique identifiers; selecting, by the device, a subset of the plurality of unauthenticated unique identification records that meet a selection criteria; segmenting, by the device, the time frame into a plurality of disjoint time intervals; determining, by the device, possible combinations of bit patterns representing the respective unauthenticated unique identifiers, wherein a length of the bit patterns equals a quantity of the time intervals and each bit of a bit pattern indicates whether a corresponding unauthenticated unique identifier has an associated unauthenticated unique identification record that meets the selection criteria for a time interval associated with the bit; determining, by the device, a total quantity of possible churn patterns for the bit patterns; determining, by the device, a total quantity of expected unauthenticated unique identifiers for all combinations of the bit patterns and the churn patterns; and determining, by the device, a ratio of unauthenticated unique identifiers to unique users based upon the total quantity of expected unauthenticated unique identifiers and the total quantity of the churn patterns.
地址 Mountain View CA US