lecture 2

1. veracity (quality)

how correct the data is, shows if we can trust the data

challenging因为易于发生，影响巨大且难以控制

2. variability

variety指same data, different object

如冰淇淋有各种不同的口味

variability指same data, different meaning

如两句相同的话在不同的时间有不同的意义

3. visibility

capture and properly present the characteristics of data

common types: charts, tables, graphs, maps, infographics, dashboards

难度体现在选择最合适的方式体现数据特征，需要结合数据特征以及目的；同时数据视图化本身也是有难度的（对于高维度的要先降维；数据本身没有结构的如区分文中积极与消极的语气，可以标注成不同的颜色；scalability可伸展性，如很多点集中在一起是如何分辨；动态数据）

4. value

value from other V's

5. in general

fundamental V's: volume, variety, velocity

characteristics/difficulties: veracity, variability

tools: visibility

objective: value

6. big data management is to server the purpose of big data analytics

7. data acquisition

application oriented: 确定什么样子的信息是问题所需要的

comprehensive: 尽可能全面的收集信息

handle data: 处理来源不同种类不同的信息

8. data storage

a) traditional way: 为structured data设计的, disk-oriented，大数据不适用

b) big data era

b.1) RDBMS -- SAP HANA

b.2) NoSQL -- HBase, Hive, MongoDB

b.3) Distributed file systems -- HDFS

9. data preparation

a) data exploration: understand your data

b) data pre-processing

data cleaning -- veracity

data integration -- variety

10. data explore

trends, correlations, outliers, statistics(mean, mode, median, standard deviation, dange: 可用来数据处理，如身高中大部分都是180，175，一个17的数据就可以被认为是dirty data)

11. data cleaning

dirty data types:

miss values/records: remove the record

invalid data; use another data as replacement

inconsistency: do additional works

duplicate: merge

outliers

12. data integration

merge data from multiple, complex and heterogenous resources to perfrom a unified view of data

13. data curation

data curation includes all the processes needed for principled and controlled data creation, maintenance, and management, together with the capacity to add value to data

数据策划包括原则性和受控数据创建，维护和管理所需的所有过程，以及为数据增值的能力