QM323 Midterm

Problems in Age of Analytics video

- there aren't enough individuals fluent in data science. - managers are not as knowledgeable about analytics, so they are unable to move past smaller scaled experiments

Parameters

-are numbers in an equation such as your objective equation -Parameters do not change when you change the choice variables

optimization models

-to make the best choice -businesses build models to help simulate what would happen as a result of each potential choice

Steps in k-Means clustering

1. Randomly partition the observations into k clusters 2. Normalize the data (i.e convert into z-scores) 3. Sort the data by cluster 4. Calculate the cluster centroids - Centroids = average values of a (normalized) variable across observations in the cluster 5. For each observation, calculate distance to each cluster centroid 6. Reassign observations to the cluster with the closest centroid 7. Repeat until there is no change in the clusters (or a specified maximum number of iterations is reached)

Decision making framework

1. find objective 2. analyze data 3. identify risks 4. communicate findings

You have a data set which includes a person's age, weight and income. You conduct a cluster analysis on all the variables that have been normalized and create two clusters. Which of these describes the centroid of Cluster 1? • 1 value - the average of the normalized age, weight, and income in cluster 1 • 1 value - the standard deviation of normalized age, weight, and income in cluster 1 • 3 values - the average of each of the normalized age, weight, and income in cluster 1 • 3 values - the standard deviation each of the normalized age, weight, and income in cluster 1 • 6 values - the average and standard deviations for each of the normalized age, weight, and income in cluster 1

3 values - the average of each of the normalized age, weight, and income incluster 1

Standardize excel function

=STANDARDIZE(X,Mean,StandardDeviation)

Absolute Reference

A cell reference that does not change when a formula is copied to a new location.

What is an influence chart?

A diagram that shows what outcomes will be generated, and which inputs influence what Ex. Revenue and costs affecting profits

Conditional formatting

An Excel feature that enables you to specify how cells that meet one or more given conditions should be displayed.

Cluster centroid

Average value of the objects contained in the cluster on all the variables in the cluster variate.

COUNTIFS

Counts the number of cells specified by a given set of conditions or criteria

COUNTIF

Counts the number of cells within a range that meet the given condition

Descriptive analytics

Defines an organization's past data

If we take the average of all the individuals estimated partworths, we always get an accurate and complete representation of the group's preferences for each attribute True False

False

True or false? When clustering, it is essential to calculate z-scores of all variables in Excel BEFORE running XLMiner to ensure results are not biased by differences in variable scale • True • False • Not sure

False

Which of these is true of k-means clustering? • It allows you to use both numerical and categorical variables for clustering • It tells you how many clusters to use • It tells you which clusters should be targeted • We can only have 2, 3 or 4 clusters It doesn't tell you which clusters are the most valuable

It doesn't tell you which clusters are the most valuable

The lift ratio of an association rule with a confidence value of 0.43 and in which the consequent occurs in 6 out of 10 total transactions has a lift ratio of 0.072 • 0.258 • 0.600 • 0.717 • 1.600

Lift = 0.43/(6/10) = 0.717 or 0.72

Break-even analysis

NPV Zero

Steps in building a good spreadsheet

Plan the structure Build the spreadsheet Review and test (and, if necessary, re-do) Communicate findings (display results in an easy to understand manner)

Steps in the decision making process

Problem recognition Information search Evaluation of alternatives Product choice

Importance

Range / (sum of ranges across all attributes) PERCENTAGE

The following elements of an influence chart can be preceded by other elements: a) Intermediate Calculations and Output b) Intermediate Calculations and Inputs c) Decision Variables and Inputs d) Decision Variables and Output e) Intermediate Calculations and Decision Variables

Recall that Inputs and Decision Variables are at the beginning of theInfluence Chart and the Objective (Output) is at the end. Intermediate Calculations are in the middle. Nothing can precede Inputs and Decision Variables. So, in the options below, only the first one is valid

object for optimization

The objective is what we are trying to maximize or minimize

Which of these is the MAIN reason to use comments for a cell: To make the spreadsheet more engaging • To document data and assumptions • To understand which cells may be referencing that cell • To link back to a different sheet

To document data and assumptions

Okcupid Article

Used k-means clustering to find women with similar answers to questions to be able to match to women of different interests and personalities

Facebook dark ads

When does ad targeting become discrimination? - Article discusses how groups were excluded from seeing particular ads for jobs or listings, depending on gender, race, etc.

You have a data set with age, weight, and income. You run cluster analysis to create 3 distinct clusters. When you plot income and weight by cluster, you find that many data points are overlapping and not clustered cleanly. Which of these is likely true: • You did not run enough iterations to identify 3 distinct clusters • The data requires you to run a different number of clusters • The inter-cluster distances must be too high • You don't necessarily expect to see clear clustering on 2 variables • Income and weight are not correlated

You don't necessarily expect to see clear clustering on 2 variables

Sensitivity Analysis

a capital budgeting tool that determines how the Npv varies as a single underlying assumption is changed

market basket analysis

a statistical technique that reveals customer behavior patterns as they purchase multiple items

In an Excel workbook, a number in a cell is calculated using 8 other cells. Something is not correct with this calculated number, and I want to see which of the 8 cells may be causing the problem. The best way that will help identify the wrong cell is: a) Linking between sheets b) Put a comment on each cell c) Trace precedents d) Trace dependents e) Start all over again

c) Trace precedents

Lift Ratio

confidence / (support of con / # of transactions)

Formula to decide who to target

cost per mailing / profit per customer

You have stored Days Per Month (value 20) worked in cell B13 and Hours per Day (value 8) in cell C13 and Hourly Wage (value $15) in cell D13. Which formula would be the BEST to calculate the total labor expense for one employee a) =20*8*15 b) =20*8*D13 c) =20*C13*D13 d) =B13*C13*D13 e) None of the above

d) =B13*C13*D13

=RANDBETWEEN(X,Y)

generates a random number equally likely to be anywhere between X and Y

Prescriptive analytics

help us identify the best course of action given a set of alternatives

Range

max partworth - min partworth

Euclidean distance

measures dissimilarity between observations on continuous variables common to normalize data into z score

A decision variable in an influence chart is a _________

oval

An objective in an influence chart is a ________

rectangle

An intermediate variable/quantity is a ________

rounded rectangle

Cluster analysis is used to

segment observations into similar groups, making data more manageable and identify outliers

Euclidean distance formula

sqrt ( (x2-x1)^2+ (y2-y1)^2 )

Confidence

support of ant + cont / support of ant

Conjoint Analysis

technique used to develop an understanding of the attributes that guide consumer preferences by having consumers compare product preferences across varying levels of evaluative criteria and expected utility

ABC analysis

the ranking of all items of inventory according to importance

predicitive analytics

uses models constructed from past data to make predictions ex) using survey data to help predict market share of new product

The technique "Association Rules" is also called (Select ALL that apply): Go to Market Analysis • Affinity Analysis • Basket of Goods Market • Market Basket Analysis • Building Lift

• Affinity Analysis • Market Basket Analysis

What is the purpose of using an error trap in your Excel workbook? • To identify a cell where you may have entered a wrong number • To identify a cell that contains an implausible number • To draw attention to a cell whose value may be of concern • All of the above

• All of the above

in a workbook, when you identify and isolate input parameters in a single location, the advantage(s) is/are (select all that apply): • Makes it easier to document assumptions behind parameters if they are in a single location • Helps avoid mistakes • Makes it easy to change the spreadsheet for sensitivity analysis • All of the above

• All of the above

Which statement is true about the K-Means algorithm? • All variables must be categorical • The output attribute must be categorical • All variables must be numeric and continuous • Variables may be either categorical or numeric • None of the above

• All variables must be numeric and continuous

Which of these is required to get meaningful regression coefficients in a conjoint analysis? Survey more than 30-40 people • Remove one level from each attribute in the regression data • Provide the same number of levels for each attribute • Include all possible product profiles in the survey • None of the above

• Remove one level from each attribute in the regression data

Which of these is the most accurate description of a cluster centroid? • The average value across all variables and clusters • The central observation in the data set • The central observation in each cluster • The average for each variable for each cluster • The median value of each variable across all clusters

• The average for each variable for each cluster

What is the purpose of an influence chart? To help define our objective • To calculate the numbers in our problem • To help analyze our model • To help design the spreadsheets / workbook

• To help design the spreadsheets / workbook

Which of these best describes the purpose of a sensitivity analysis? • To understand which variables are likely to change over time • To understand the impact changing one variable will have on the outcome of interest • To calculate the values of each variable that will maximize the outcome of interest • To determine how many units will be sold based on segment size • All of the above

• To understand the impact changing one variable will have on the outcome of interest

The K-Means algorithm terminates when... • a user-defined minimum value for the summation of squared error differences between instances and their corresponding cluster center is seen • the number of instances in each cluster for the current iteration is identical to the number of instances in each cluster of the previous iteration • there are three clusters • the number of clusters formed for the current iteration is identical to the number of clusters formed in the previous iteration • the cluster centers for the current iteration are identical to the cluster centers for the previous iteration

• the cluster centers for the current iteration are identical to the cluster centers for the previous iteration