COMP9313 Week9a-0

https://drive.google.com/drive/folders/13_vsxSIEU9TDg1TCjYEwOidh0x3dU6es

https://www.cse.unsw.edu.au/~cs9313/20T2/slides/L8.pdf

 

Mining Data Streams

 

1.  Data Streams

　　1)•Stream Management is important when the input rate is controlled externally  输入率由外部控制

　　2)We can think of the data as infinite and nonstationary (the distribution changes over time) 数据是无限，且一直变化的

 

2. DBMS VS Data Stream 

　　1) random access is expensive(on the disk) - single scan algorithm 

 

3. Sampling from a Data Stream

　　1) 取10%（fixed proportion）的数据

　　　　1） Naive Approach: 随机取百 分之十

　　　　2）

　　2）取fixed size的数据