C207 Final Exam Study Notes Understanding Analytics and Data Quality | Accurate & Verified Answers to Pass Actual Exam

C207 Final Exam Study Notes Understanding Analytics and Data Quality | Accurate & Verified Answers to

Pass Actual Exam
C207 Cohort 1:
I want to buy a new car/ house: In business – where do I find pertinent information
oWe do analytics BEFORE then measure our results
oThings that can be measured between vehicles to make best decision: comparable
FACT BASED information for decision to build TRUST
Makes value of your decisions higher (employees, customers, suppliers)
Make accurate predictions – to reduce RISK
Types of Analytics: Ask yourself 2 questions: Am I predicting? Am I optimizing?
oDescriptive:
Past Data Only: “car prices up 2% in past year”
NO prediction: NO optimizing
oPredictive:
Past to predict future: “based on past 10 years, car prices expected to raise 2% next
year”
YES Prediction: NO optimizing
oPrescriptive:
Past to predict future and Optimizing: “By increasing electric charging stations by 7%,
electric car sales are expected to make a 5% increase next year”
YES predicting: YES Optimizing
Data Quality: Errors in data
oOmission: (find easily by sorting by columns in Excel)
oOut of Range: (find easily by sorting by columns in Excel)
oOutlier is NOT AN ERROR

Think about what you’re left with. Is it gone?
oSystematic Errors: Error will not fix itself (skew the data)
oRandom Error: Error will fix itself (with lots of data)
Reliable vs. Valid
oReliable: Consistent and repeatable / a measure of the instrument
Instrument: thermometer – bad reading, do it again
Using reliable instrument = valid data results
oValid: Measures what is intended to me measured / does test score represent ability?
BIAS:
oMeasurement Bias:
Representative sample: every member of group (population) has equal opportunity to
be selected
At least 30?
Randomly selected (to eliminate bias)
oInformation Bias: ignoring the purpose of the information collected
Questions / information not relevant to the goal of the survey
Record everything, weed out irrelevance later (don’t decide up front)
Non-truthful answers: try to get the bias out of the way up front

, Anonymous makes more truthful
BIG DATA:
oWHAT? Both structured and unstructured data in such large volume that it’s difÏcult to process
using traditional database and software techniques.
Structured: grocery store checkout
Unstructured: don’t fit in rows, columns (social media, email, photos, file notes)
oWhere? Servers in a large data warehouse. (3rd party)
oWhy? Used to encourage buying behavior. (increase customer base and BUY)
Data-mining used to discover patterns in large data sets.



1.04 Rise of Analytics (turning information into insight and developing conclusive fact-based strategies
to gain a competitive edge) An analysis is the key to unlocking the value of big data
oStatistics: interpretation of numerical facts or data through theories of probability
oAnalytics: analysis of meaningful patterns in data
Descriptive (diagnostic): depict and describe what is studied- what has already
happened. (used extensively in business)
Graphical analysis: charts
Discovery of patterns: numbers
Predictive: Data from past predicts future or the impact of one variable on another
Models: trend analysis, regression
Simulations
Prescriptive: includes experimental design to suggest course of action
Optimization Models
Simulation
Decision Analysis: decision trees. Several alternatives with uncertain future
events.
Management Science: the study of model building, optimization, and decision making

, 1.07 Models of quantitative Decision making: Davenport-Kim 3-stage model
oFraming the problem (define problem, types of analysis, what/how to collect data)
Problem recognition
Identify stakeholders: who are you reporting to?
Focus on Decisions: what decisions will be made from analysis? Worth it?
Identify type of story you’re going to tell: start thinking
Determine scope of problem: don’t get too small yet: consider all
Get specific: focus on narrow set of data you will analyze
Review of previous findings: help you structure your analysis and give ideas for
storytelling when analysis is complete
oSolving the problem (type of data, data collection, data error/analysis technique)
Modeling: simplified representation meant to solve a particular problem
Study sales by demographics:
Data collection: data gathered by primary or secondary sources
Structured: fits easily in spreadsheet
Unstructured: text, images, clicks: need quantified before analysis
Data Analysis: find patterns in data that can be explained using more sophisticated
statistical techniques. Level of analysis depends on story you’re telling
oCommunicating results: extremely important to get your results acted upon/ make
recommendation.
MAKE INFORMED DECISIONS
Succinctly, efÏciently portray the meaningful results
Match level of background: scientist need proof, not just results
Don’t “talk down” to less scientific managers
1.14 Video: The use of research
oMake sense of patterns
oSurveys, observations, predictions
oResearch informs decisions and actions
oResearch design and execution are paramount
1.15 Research Design: Two main types of design
oObservational Studies: “quasi-experimental” studies
Used when impossible or impractical to control the conditions of the study
“Mystery shopping’
Variables monitored in natural setÝng: what’s observed isn’t artificially
constrained
Best are:
Forward-looking
Prospective
Focus on random group
Issue: weaker in statistical inference
Prospective cohort study: observes people going forward in time after entry in the study