2 Week Flashcards
Analytics with Ba
- Recognize a problem
- Define the problem
- Structure the problem
- Analyze the problem
- Interpret results
- Implement the solution
Realizing the problem
Findinf sing of issues
Define the problem
clearly define the problem, what is part and what is not?
Structure the problem
stating goals and objectives, charactherizing the possible decisions
Analyze the problem
Analytics plays a major role, involves experimentation and modelling process.
Interpret reults and make decision
What do the results found by the model mean for the application?
implement the solution
Translate the results of the model back to the real world
data
- Raw facts
- No context
- Just numbes and text
Information
- Data with context
- Processed data
- Value added to data
Knowledge
- Knowing what information is required
* Knowing what the information means
Data source (internal)
- Annual reports
- Accounting Audits
- Financial profitability
- Operation managment performance
- Human resource measurements
Data source(External)
*Economic trends
*Marketing research
*
New development
- Web behavour
- Social Media
- IOT
Big Data
refer to massive amounts of business data from a wide variety of sources, much of wich is avalaible in real time, and much of which is uncertain or unpredictable. IBM calls the caractheristics volumen, variety,
Database
A collection of related tables containing records on peoplem places or things (extract with SQL)
Data set
A collection of data (often a single “spread sheet” or data mining table)
Discrete (Type of data)
derived from counting something (Ex. Is the deliver in time?)
Continous (Type of data)
based on a continous scale of measurement (any metrics involving money, lenght, time)
Measurment scales (Categorical (nominal data))
sorted into categories according to specified characteristics [Equality: IAre the colors the same?]
Measurment scales(Ordinal data)
can be ordered and ranked according to relationship to another [sort: Is this value larger or better?]
Measurment scales (Interval data)
there is order and the differences between the values is meaningful [Temperatu]
Measurment scales (ratio data)
has all the properties of an internval variable [prices]
Reliability
data accurate and consistent
validity
data measures what is supose to be measured
Experiment
Method for testing differen assumptions by trial and error under conditions constructed and controlled. Independed variables are allow to change and the effects are analyzed
“Test and learn approach”
take action with one group then take action with another group and check
- Focus on individuals and think short term (immidate response )
- Keep it simple
- start with prof-of concept test
- when results come in, slice the data
- try out-of-the-box thinking
- Measure everything that matter.
- Look for natural experiments
Tools
data base queries and analysis:
- spreadsheets
- Data manipulation
- Data visualization
- Dashboard to report KPM
Descriptive analytics
- Data source & types
- Data wharehouse
- Ghapical represatation
- Data visualiztion
- Analytics using SW
Predictive analysys
- Simple linear regression
- Multiple regrassion model
- Non-linear refression
- Time series forecasting
- Simulation models
Model
an abstraction or represatin of a real system
*Capture most importan featured
Predcitve (Category) ANOMALY
something out of the common
Predcitve (Category) ASSOCIATION LEARNING
classification and cluster
prescriptive analysis
- Optimization model
- Simulation model
- Decision analysis
- spreadsheet models