Please enable JavaScript.
Coggle requires JavaScript to display documents.
Big Data and Analytics (sensor data (sensors don't always tell the…
Big Data and Analytics
Business Analytics in Context of Big Data
the core of BD
BI
decision support
analytics
four dimensions
volume
velocity
variety
veracity
Big Data Analytics Framework
data storage
Hadoop cluster
data preparation
ETL
data cleasing
data source
structured data(the majority)
semi-structured data(used for analytics)
unstructured data(from external sources)
analysis
querying data
descriptive analytics
exploratory analytics
predictive analytics(emerging field)
analytics sandbox
data access & usage
casual users
executives
managers
front-line workers
powerful users
business analysts
data scientists
analytical modelers
research directions
New Analytics Capabilities
four lessons
leverage process focus
foster data awareness
importance of data access
adopt agile develpoment practices
leaning iteratively
identify further opprtunities
from isolated to open platforms
Analytics Ecosystem
organization
technology
business
the project
identify business value oppotunities
vision and goals
6 steps legacy process
goals
faster long-term process
incresas accuracy
increase flexibility
improve reproducibility
address data access challenges
common data repositories
data harmonization problems
more than one data source
open analytics platform(the R)
develop and implement applications
evaluation
maximizing the value from business analytics
two themes
speed to insight
transform raw data into usable information
practices
automation
business requirements
business analytics products/services
reuse
visually appealing software
rich media support
mobility
user engagement(an interactive way)
pervasive use
recommendation
optimized ecosystem
data standart
invest in business-savvy IT staff
encourage user-intensive development
embrace new technology
two important roles
graphic designer
app developer
the values
transctional
informational
strategic
sensor data
sensors sometimes lie
sensors don't always tell the whole truth
how and where sensor data has been filtered and summarized
"multi-genre" analytics and additional data
prepare data
raw sensor data
cleansed sensor data
event detection
labelled sensor data
event association
don't measure the quantity of interest directly
limited value
data integration