(此文为学习笔记，课程来自Bigdata University：http://bigdatauniversity.com.cn/courses/BigDataUniversity/PA0101/2016_06/courseware/c4323451afcd4b05946917efc8fc86f5/be5f0606db034b559b014e87ab62e418/)
Why we do data mining?
Market Context.
Analytics Drive Decision-Making.
Information age: terabytes and petabytes of data available. How do we consume this data, translate it into information and make it usable?
What is data mining?
Process of discovering insights, patterns and relationships from large amounts of data.
What knowledge can be extracted?
Descriptive: What has happened and why did it happen?
Predictive: What is likely to happen next
What can we learn?
- Association Rules: Rules that indicate relationships. For example, people who buy diapers also buy beer.
- Classification: Finding a model that describes the data and classifies it to a set of categories. For example, people who drink and drive are more likely to have higher insurance rates.
- Segmentation: Grouping objects by similarity. For example, prospective customers are broken up into clusters of suburban families with children, single college students, urban empty nesters.
The key things:
Allow us to uncover inside information.
SPSS modeler: CRISP-DM( Cross-Industry Process for Data Mining)
- Iterative process
- Business Understanding
- Data Understanding
- Data Preparation
- Modeling
- Evaluation
- Deployment
Software: IBM SPSS Modeler