Questions Flashcards
What is a model? (Generic Definition)
A representation of a thing, system or theory, such that characteristics of the original can be examined or changed
What is a model? (Business definition)
A logical computer program that shows interactive relationships between dynamic variables, resulting in output to anwer management questions
What is the difference between Model and Analysis
In analysis you follow detailed preparatory steps to see output.
In a model, whenever input or assumption change, output automatically changes.
What are the two definitions for a model?
Generic and Business
4 steps business evolution
1 - Bureaucracy
2 - Process Reengineer
3 - Data Analytics
4 - Robot Advisory Machines
What does bureaucracy step consists of?
Humans did everything, there was a full process line mindset
What does Process Reengineer step consists of?
Re-design processes and optimize based on value
What does Data Analysis step consists of? (Business Evolution)
Data-driven decision-making process. Decisions based on KPI. Automation of repetitive tasks.
What does Robot Advisory Machines step consists of?
Operative business work developed by Robotic Process Automation / Machine Learning
Why use a model? (Name the 3 factors)
Accuracy, Efficiency and Interactivity
What is Accuracy? (Why use a model?)
Accuracy means that everything can be correctly represented in the form of a model.
What is Efficiency? (Why use a model?)
Time saving on repetitive tasks. Simplifying business problems by breaking them down.
Document data sources, assumptions and methodologies in one place (nothing left unanswered).
What is Interactivity? (Why use a model?)
“What if” analyses to check solutions sensitivity to changes.
Produce content for presentations, or be a presentation tool itself.
Our journey steps?
Data Collection; Data Preparation; Data Visualization; Data Analysis; Insight Communication; Take Action
What is Traditional BI?
Measures past performance and guides business planning. (E.g. Reporting (KPIs, metrics), Dashboards, Scorecards, Querying)
What does Traditional BI respond to?
What happened? Why did it happen?
What is Advanced Analytics and Data Science?
Data Mining; Statistical Analysis; Predictive Modelling; Big Data Analytics; Simulation; Optimization; Machine Learning
What questions do Advanced Analytics and Data Science respond to?
What will happen? How can we make it happen?
Analytics in Traditional BI
Descriptive Analytics and Diagnostic Analytics
Analytics in Advanced Analytics and Data Science
Predictive Analytics and Prescriptive Analytics
Modelling Overview
Variable Assumptions (UI), Fixed Assumptions, Data into Engine. Engine gives us Output (UI)
Three steps to build a model
Structuring and Design; Construction; Use and Care
What is done in Structuring and Design?
Determine model’s purpose
Define output
Identify required data
Define flow and content of major sections
Map the physical layout
Structuring and Design advantages
High ROI
Over 10 times return on time spent
Can be a deliverable itself
What is done in Construction? (Build a model)
Establish assumptions
Input required data
Format model for easiest use based on established structure
Test for errors
Construction advantages (How to build a model)
With the proper Strucure and Design, the model will write itself
What is done in Use and Care? (How to build a model)
Test logic and limits
Update and modify the model as needed
Document the model
What are the tests in Testing Policy?
Error-testing
Sanity-testing
Sensitivity and stress testing
What does Error testing consist of?
Use set of known inputs to test if the model produces expected results
Test for internal consistency
Parts should sum to the whole (e.g. customers, geographies)
What does Sanity testing consists of?
Vary inputs in a controlled manner to test its directional logic (e.g.: unit costs decrease 15% for each doubling in volume, and uc increase)
Check assumptions and methodologies validity (can revenue grow 20% annually while cost grows 5%?)
Check output validity (Are ROEs different from peer businesses?)
What does sensitivity and stress testing consist of?
Sensitivity of output to small changes in key inputs
Test model against outer ranges of inputs to understand models limitation (test model when you “zero-out” certain inputs)
Credibility can quickly disappear if the model produces “ridiculous” results outside application ranges
What does Data Collection consists of?
Collect raw data on business operations from different sources. Much of the data is generated automatically. Strategy may be required to capture certain data and address critical business questions.
What does Data Preparation consists of?
Clean, combine and format data for reporting and analysis purposes. Accurate and consistent data is important to obtain valuable insights from what is collected.
What does Data Visualization consists of?
To monitor business performance, data must be visualized in reports and dashboards. By sharing summarized information throughout an organization, managers and employees will be aware of business performance in different aspects.
What does Data Analysis consists of?
To gain insights, people will need to explore the data for potential issues or opportunities. An iterative process of data discovery can help organizations unlock insights.
What does Insight Communication consists of?
Insights must be communicated effectively (to ensure right decisions and actions). Data storytelling using engaging narratives and clear visuals to explain insights opens audience’s minds to nem and better possibilities.
What does Take Action consists of?
Which insights should be pursued and implement necessary changes. (Test can be deployed first, to verify results). Regardless, results should be assessed after each change, and learned from.
What is a Programming Language?
A set of rules that converts strings or graphicar program elements (visual programming languages), to various kinds of machine code output. Are one kind of computer language and are used to implement algorithms.
What is an Algorithm?
A finite sequence of rigorous instructions. Used to solve a class of specific problems or to perform a computation. Used to perform calculations and data processing. Advanced Algorithms can use logic to divert code execution through various routes (automated decision making)
What is a Computer Program?
A sequence or set of instructions in a programming language for a computer to execute. One component of software.
What is Data Visualization? (Non-journey part)
Graphical representation of information and data.
Data visualization tools provide a way to understand trends, outliers and partterns in data.
Great way to present data to non-technical audiences.
Regarding Big Data, data visualization tools are essential to analyse massive amounts of information and produce data-driven decisions.
Common reading patterns?
F-pattern
Z-pattern
Most common reading pattern for big amount of content?
F-Pattern
Z-Pattern is typical for?
Dashboards that are not content-heavy
How do we guide user’s attention?
Take advantage of reading patterns and dashboard form.
What is High-Frequency Trading?
Method of trading that used powerful computer programs to transact a large number of orders in fractions of a second. Uses complex algorithms to analyse multiple markets and execute orders based on market conditions.
What is a stock exchange?
Physical or digital place where investors buy and sell stock or shares in publicly traded companies. Share prices are driven by supply and demand.
What is a business model?
Describes how an organization creates, delivers and captures value, in various contexts.
Business models need to account for…
Competitors. Businesses best decisions are based on what is best for the company, taking into consideration how others will act and decide.
How can we characterize a network?
Degree distributions; Average Distances; Diameter; Shortest Path; Transitivity; Clustering; Community Structures; Robustness
What are Hubs?
Nodes with a huge number of connections
What is the difference between variable and fixed assumptions, in the context of a model?
Fixed assumptions are a part of the model, meaning we always assume they hold true, if they stop doing so, than our model stops making sense. Variable assumptions, as the name says, are variable, they can change and the model adapts to them. Often are introduced by the user as a part of UI.
What is the difference between printing and returning, inside a function?
Print just shows the user a string representing what is going on. The computer cannot make use of a print.
Return is how a function gives back a value. This value can or cannot be seen by the user, but it can be used in further functions.
