Targeted multi-dimension data extraction for real-time analysis

摘要

Methods and systems for extracting targeted data for real-time reporting are discussed. In an example, a system can include a data store, a server, and a denormalized database. The data store can maintain data created by an application. The server can be communicatively coupled to the data store. The server can include a data extraction module and a data compression module. The data extraction module can extract a subset of the data stored in the data store according to an extraction scheme. The data compression module can compress the extracted subset of the data into a set of aggregated key value pairs. The denormalized database can store the aggregated key value pairs.

主权项

1. A method comprising:
storing transaction data generated by an online application in a data store to maintain the transactional data, the transactional data comprising sessions of users with each session comprising a start defined by the corresponding user accessing the online application and an end defined by one from the following group: the corresponding user logging out of the online application, the corresponding user navigating away from the online application, and the corresponding user becoming inactive for a pre-determined amount of time; extracting, using one or more processors communicatively coupled to the data store, data attributed to the online application from the transactional data; extracting tracked operations from the extracted data attributed to each session of the sessions, the tracked operations attributed to each session are interactions of the corresponding user with the online application; identifying within the tracked operations one or more successful operations; attributing each of the identified one or more successful operations to at least two of the tracked operations, the at least two of the tracked operations are from the corresponding session of the sessions, and the at least two of the tracked operations and the corresponding successful operation are a chain of operations with the at least two of the tracked operations being ancestors of the corresponding successful operation; dividing an attribution score for the corresponding successful operation of the one or more successful operations between the at least two of the tracked operations; and causing a report to be displayed on a computer display, the report indicating at least the attribution score for the at least two of the tracked operations.