Simulation DATA-SCIENCE-OPTIMIZE Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A machine learning engineer has developed a predictive model for equipment failure and wants to validate its performance under extreme operating conditions that are rare in the training data. Which approach is most appropriate?

   • AUse cross-validation on the existing training data to get an average performance metric.
   • BUse bootstrapping to resample the training data and refit the model to each resample.
   • CUse Monte Carlo simulation to generate synthetic data representing extreme conditions, then evaluate the model's performance on this simulated data.
   • DUse a holdout test set that contains only normal operating conditions.
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A data scientist is building a Monte Carlo simulation to model the total number of defects in a manufacturing process. The number of defects per batch follows a Poisson distribution with mean 5. The scientist needs to generate random numbers from this distribution. Which approach is most appropriate?

   • AUse a random number generator that directly samples from a Poisson distribution with λ = 5.
   • BGenerate a random number from an exponential distribution with mean 0.2 and round it to the nearest integer.
   • CGenerate 5 random numbers from a uniform distribution between 0 and 1 and sum them.
   • DGenerate a random number from a normal distribution with mean 5 and standard deviation √5, then round to the nearest integer.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A project manager is evaluating the net present value (NPV) of a new project. The NPV depends on several uncertain inputs: initial investment, annual cash flow, and discount rate. The manager wants to know which input has the greatest impact on NPV variability. Which approach is most appropriate?

   • AVary each input one at a time by a fixed percentage and observe the change in NPV.
   • BSet all inputs to their most likely values and compute a single NPV.
   • CRun a Monte Carlo simulation that varies all inputs according to their probability distributions, then compute the correlation between each input and the NPV across simulations.
   • DUse bootstrapping on historical project data to estimate the distribution of NPV.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   What is the role of a random number generator (RNG) in creating simulated datasets?

   • ATo guarantee that the simulated dataset exactly matches real-world data
   • BTo generate the same sequence of numbers every time without any seed
   • CTo eliminate the need for specifying input distributions
   • DTo produce numbers that follow a specified probability distribution
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A manufacturing company uses a simulation model to predict production yield. The model has three uncertain inputs: temperature (normally distributed), pressure (uniformly distributed), and material quality (discrete with three levels). The company wants to identify which input has the greatest impact on yield variability, but the inputs are correlated. Which approach is most appropriate?

   • AUse a Monte Carlo simulation that samples from the joint distribution of the inputs (accounting for correlations) and then compute the partial rank correlation coefficient (PRCC) between each input and yield.
   • BPerform a one-at-a-time sensitivity analysis by varying each input across its range while holding others at baseline values.
   • CRun a Monte Carlo simulation that varies all inputs independently and compute the Pearson correlation between each input and yield.
   • DUse bootstrapping on historical production data to estimate the variability of yield.
   Reveal Solution Discussion Select an answer first