Data Engineering for MADS: Weekly Quiz 1, 5 & 6

What is according to Arnoud de Munnik from Wehkamp.nl the challenge of the domain 'Math & Statistical Knowledge'?


Building models that add value to the business


Building models with the shortest runtime


Building models with the best performance


Building models that will work

What are 'measures' (or KPI's) that Wehkamp.nl often uses according to Arnoud de Munnik?


Product margin, Conversion, Returns, Top 10 ssearch terms


Product impressions, Conversion, Returns, Top 10 search terms


Product impressions, Conversion, Returns, Customer Lifetime ValueC


Product margin, Conversion, Returns, Customer Lifetime Value

Locating and identifying data elements in source files is called: Parsing, Extracting, Querying OR Consolidating?

Combining the beach wear categories labelled 'boxer shorts' and 'boxer-shorts' in one category is a form of... Matching, Correcting, Consolidating OR Standardizing?

Which of the items below is not a requirement for tidy data? Variables in columns, Observations in rows, Column headers are values OR One type per data set.

When respondents are reluctant to share low income values, missing values for the income variable are considered as: random missing values, MAR, MCAR or MNAR?

Consider the case where elderly people are less likely to report their income. Which of the options below is correct?
(1) Missing observations for income are MAR, if age is included in the analysis
(2) Missing observations for income are MAR, if age is not included in the analysis
(3) Missing observations for income are MNAR, if age is included in the analysis
(4) Missing observations for income are MNAR, if age is not included in the analysis

Consider the following statements about MAR methods:

A. MAR methods are mostly better than traditional/naive methods

B. Methods for MNAR are always better than MAR methods

Which of the statements below about listwise deletion is not true?


Listwise deletion is a reasonable approach if not more than 5% of the data points is lost


Listwise deletion may lead to biased results


Listwise deletion always leads to loss of statistical power


Listwise deletion is the preferred option by most Data Scientists

What is the most important advantage of multiple imputation?


Multiple imputation filters out errors from various sources

Averaging across multiple imputations leads to more accurate estimation results

By repeating the imputations, the results converge faster

Multiple imputation allow for errors from various sources to affect the imputation

Which of the statements about the Mahalanobis distance is not true?


Assessing the Mahalanobis distance is the multivariate equivalent of the manner in which outliers are identified in a boxplot


Values of the Mahanalobis distance should be assessed with a t-distribution


Is a distance measure that represents the deviation of a data point to the overall center of the data


The Mahanalobis distance takes the relationship between the variables in the analysis into account

Consider the sequence of numbers: 1, 3, 6, 12, 14, 20, 24, 36, 73, 100. Which of the statements below is true?


The 20% trimmed mean of this sequence equals 18.67


The 20% trimmed mean of this sequence equals 23.5


The 20% trimmed mean of this sequence equals 26.4


The 20% trimmed mean of this sequence equals 28.9

Combining several sources into one model to solve a defined problem is called...

Problem Solving

Collateral Catch

Data Exploitation

Data Mining

There are different types of analysis with different levels of complexity. Which order, from low to high complexity, is correct?


Descriptive, Diagnostic, Predictive, Prescriptive


Diagnostic, Predictive, Descriptive and Prescriptive


Diagnostic, Descriptive, Predictive, Prescriptive


Prescriptive, Predictive, Diagnostic, Descriptive

Which of the following statements is true?


Data Science is a subset of Machine Learning


Machine Learning is a subset of Artificial Intelligence


Deep Learning is a subset of Data Science


Machine Learning is a subset of Deep Learning

For descriptive analytics you can make use of reports, also known as key figures. Which plots are often used in reporting analytics?


Scatterplot and line graph


Boxplot and histogram


Histogram and line graph


Scatterplot and boxplot

Which test do you need if you want to compare a one-to-one relation between a numerical KPI and a categorical potential driver with 4 levels?


ANOVA


T-test


Chi-square-test


Wilcoxon rank test

Levene's test is


A test for multicollinearity


A test for homogeneity of means


A test for homogeneity of variances


A test for normality

In a decile analysis the customers are divided in 10 equal-sized groups. How are these groups created?


Based on a customer characteristic, e.g. zip code or age


Based on their customer number


Based on the ordered KPI (from high to low)


Randomly in 10 groups

Which insight can a migration analysis give?


To understand changes in aggregate sales (or another important KPI)


To understand how customers behave (e.g. which product(combinations) do they have over time)


To understand the status of the customers (e.g. churn)


All of the above

Which of the following statements is correct?


Cluster analysis is the same as a regression analysis


Cluster analysis is a segmentation technique


Cluster analysis can be done using a principal components analysis


Cluster analysis can work with a large number of variables

What are often used value drivers in a like-4-like analysis?


Recency, Frequency, Monetary value


Revenue, Service Costs, Retention


Response rate, Conversion rate


Inflow, Upsell, Stable, Downsell, Churn