Answer-first summary for fast verification
Answer: ML does not test a specific hypothesis but lets the data decide on the best features to be included in the model; traditional techniques decide on both the model and the variables.
**B is correct.** For traditional techniques, the analyst decides on the model and the variables to include and infer from the results whether the data support the pre-specified theory. ML, on the other hand, lets the data decide on the best features to include in a model and does not test a particular hypothesis. **A is incorrect.** This refers to traditional techniques. **C is incorrect.** The role of computer algorithms in traditional econometric and statistical techniques is generally limited to estimating the parameters and checking for statistical significance. **D is incorrect.** ML is mainly concerned with performance measures focused on the accuracy of predictions or classifications; statistical significance, goodness of fit, and residual diagnostics are relevant for traditional techniques.
Author: LeetQuiz .
Ultimate access to all questions.
No comments yet.
A data scientist is presenting to a group of quantitative analysts about the benefits of using machine learning (ML) in developing trading strategies and constructing portfolios. The data scientist compares ML techniques to the classical statistics and econometrics currently employed by the analysts when creating forecasting models. Which of the following correctly differentiates ML from traditional econometric and statistical techniques?
A
ML assumes that the data-generating process can be approximated based on economic theory, but traditional statistical techniques do not make this assumption.
B
ML does not test a specific hypothesis but lets the data decide on the best features to be included in the model; traditional techniques decide on both the model and the variables.
C
Unlike in ML, the role of computer algorithms in traditional econometric and statistical techniques generally goes beyond estimating the parameters and checking for statistical significance.
D
Analyses of goodness of fit and residual diagnostics are both relevant for ML but not for the traditional techniques, as they are mainly concerned with the accuracy of forecasts.