Unit 1: Introduction to Data Science - Practice Quiz

1 What is a primary reason for learning data science?

2 What does the data analytics life cycle describe?

3 What is a key activity during data discovery?

4 Which task is commonly performed during data preparation?

5 What is selected during model planning?

6 What happens during model building?

7 What is the main purpose of communicating results?

8 What does operationalization usually involve?

9 Which is a recognized type of data analysis?

10 Which question does descriptive analysis primarily answer?

11 Which question does diagnostic analysis primarily answer?

12 What is the purpose of predictive analysis?

13 Which question does prescriptive analysis primarily answer?

14 Which list contains the four common types of data analytics?

15 How are missing values usually handled in an analytics project?

16 Which stage commonly follows communicating the results in the data analytics life cycle?

17 Which tool is especially useful for communicating analytical findings?

18 Which item should be clarified during data discovery?

19 Why is a model tested after it is trained?

20 How can data science support an organization?

21 An online retailer wants to increase repeat purchases using its transaction and browsing data. Which data science application best supports this goal?

22 A team has defined its business problem and identified relevant data sources. What should it generally do next in the data analytics life cycle?

23 During data discovery, a hospital team finds that patient identifiers differ across two databases. What is the most appropriate initial response?

24 A dataset contains missing ages, and the analyst plans to evaluate a predictive model on a test set. Which approach best avoids data leakage?

25 A bank wants to predict whether each loan applicant will default within one year. Which modeling task should be planned?

26 Only 2% of transactions in a dataset are fraudulent. Which evaluation measure is generally more informative than accuracy for model planning?

27 A model performs very well on training data but poorly on validation data. Which action is most likely to improve its generalization?

28 An analyst must explain a sales forecasting model to regional managers. Which presentation is most effective?

29 A deployed demand model becomes less accurate after customer behavior changes. Which operational practice would detect this problem?

30 Website conversions fell last month. An analyst compares traffic sources, page errors, and device categories to determine the cause. What type of analysis is this?

31 A manager asks for a summary of last quarter's revenue by region and product category. Which output best addresses the request?

32 A factory's defect rate increased from 3% to 7%. Which method is most suitable for investigating the increase?

33 A telecommunications company uses past customer behavior to estimate which subscribers may cancel next month. What is the primary output?

34 A delivery company has predicted tomorrow's package volumes. It must now assign trucks while minimizing cost and meeting capacity limits. Which approach is appropriate?

35 Which sequence correctly progresses from understanding past performance to recommending a future action?

36 A customer dataset has a nominal feature called payment_method with values such as card, cash, and transfer. Which transformation is suitable for many machine-learning models?

37 During model building, a team discovers that inconsistent measurement units are reducing accuracy. What should the team do?

38 An analyst tunes several models using the validation set. How should the analyst obtain a less biased estimate of the selected model's final performance?

39 A model estimates that a marketing campaign will increase conversion by 4%, with a wide confidence interval. How should this be communicated?

40 A company plans to deploy a new recommendation model gradually. Which strategy best reduces operational risk?

41 A retailer has accurate demand forecasts but still experiences frequent stockouts and excessive holding costs. Which capability most strongly explains why learning data science remains valuable beyond producing predictions?

42 During model validation, a team discovers that its strongest predictor is recorded only after the outcome occurs. What is the most appropriate life-cycle response?

43 A hospital wants to predict readmission at the instant a patient is discharged. Which discovery finding should cause the greatest concern about a candidate variable?

44 A numerical feature has missing values, and the dataset will be evaluated with five-fold cross-validation. Which preparation procedure avoids validation leakage?

45 A model will predict equipment failures from daily sensor readings, and performance must reflect predictions on future machines and future dates. Which validation design is most defensible?

46 Only 0.2% of transactions are fraudulent, investigators can review 500 alerts daily, and false negatives are costly. Which primary evaluation design best matches deployment?

47 After tuning 200 hyperparameter configurations against the same validation set, a team reports the best validation score as its expected production performance. What is the best corrective action?

48 Each regional branch shows a higher conversion rate under campaign B than campaign A, yet the company-wide conversion rate is lower under B. What should an analyst communicate first?

49 A credit model's input distributions remain stable, but default rates increase and predicted probabilities become systematically too low. Which monitoring conclusion is best supported?

50 An analyst first quantifies last quarter's churn, then investigates its drivers, forecasts next quarter's churn, and finally allocates retention offers under a fixed budget. Which sequence correctly classifies the analyses?

51 Department X has 20 employees with an average salary of , while department Y has 80 employees with an average salary of . What is the correct company-wide average salary?

52 Sales fell immediately after a website redesign. A diagnostic analysis finds that mobile traffic also shifted toward a lower-converting country during the same week. Which conclusion is most defensible?

53 Two binary classifiers have equal ROC-AUC, but model P is well calibrated and model Q is not. Decisions use predicted probabilities to calculate expected monetary loss. Which statement is correct?

54 A delivery optimizer minimizes expected travel time but repeatedly assigns routes that exceed driver-hour regulations. What is the most appropriate correction?

55 A factory dashboard reports failure counts, a root-cause model links failures to overheating, a forecast estimates tomorrow's failures, and an optimizer schedules maintenance. Which component can recommend an action while directly considering resource constraints?

56 After deployment, users consistently override a model because one recommended action is operationally impossible. Which response best reflects an iterative analytics life cycle?

57 A categorical feature receives previously unseen values after deployment. Which training-time design most robustly prevents scoring failures without using future data?

58 For a calibrated binary classifier, a false positive costs and a false negative costs , with zero cost for correct decisions. At what probability threshold should a case be classified as positive?

59 A forecast gives next month's revenue as million with a 95% prediction interval of – million. Which statement communicates the result most appropriately?

60 A new fraud model performs better offline, but labels arrive after 60 days and a faulty release could block legitimate customers. Which rollout strategy best controls operational risk?