status

The beta version of loading click stream has been released.
The DDL for database is designed.
next step
will wait the requirment and do some pre-computing.
Also, the time is still unsolved yet.

Click flow analysis

Currently, about 30G of click records are logged. About 80M rows of records. For the first version, We'll get the basic PV PU etc.
We use the perl script to do the job for now.

And we find a good module called module bigint
Unlike integer, this pragma creates integer constants that are only limited in their size by the available memory and CPU time.

Since we can use the database to do calculation. The script will only nomolize the records and then load into the database.

We can defined the format of to be loaded data now.

current status

For now. I'm working on the click flow staff. We have no privilege to code the script of logging. We can only make use of the log files. I have developped the scripts. Waiting for the measurements now. The model is quit useful though. Many of them can be imply, some of them cannot. Will need the data of the real data to test the performance with vast number of customber. And we have to analysis the customers seperatly.