Test
Module 4 Review
Crea una cuenta gratis para terminar este test
Guarda tu progreso y consulta tu resultado final.
¿Ya tienes cuenta? Inicia sesión
Preguntas del test (16)
1There are three model essentials of Decision Tree, except
- AGrowing a tree
- BGenerating prediction rules
- CSplitting the tree
- DPruning the tree
2In optimizing Decision Tree model complexity, there are three main aspects, which are
- ADecision, Rankings, Estimates
- BDecision, Estimates, Validation
- CDecision, Validation, Implementation
- DEstimates, Validation, Implementation
3In a Decision Tree, which of the following is the measure of purity in splitting, and what is its value for a pure population split?
- AGini index; 1
- BP-value; 1
- CP-value; -1
- DGini index; 0.5
4Which of the following Split is better in a Decision Tree?
- AA split with an entropy value of 0
- BA split with an entropy value of 1
- CA split with an entropy value of 0.5
- DA split with a Gini index of 0
5Which of the following is the best example of Redundancy while choosing inputs for a predictive model?
X1 & X2 – Input Variables
Y – Predictor Variable- AX1 has the same information as X2 to predict Y
- BX2 has a more impact on Y than X1
- CX1 has no same information as X2 to predict Y
- DX1 and X2 cannot predict Y respectively
6A good split of nodes in a decision tree will do the following -
- AMaximize Gini index
- BMaximize Entropy
- C2nd correct option- Minimize Entropy
- DMinimize Gini Index
7Which of the following is true with regards to predictive modelling using statistical approach -
- AStatistical significance via p-value is the dominant philosophy
- BPrediction is more important than explanation
- CRequires variable selection methods to choose from many variables
- DModel performance is the dominant philosophy
8What do we do to improve the result when the main aim for prediction is ranking?
- AMaximize concordance
- BMinimize squared error
- CMaximize accuracy
- DMinimize misclassification
9Which of these is not a property of decision trees?
- ATrees require you to assume linear relationships between X’s and Ys
- BTrees tend to favor X’s with many potential split points
- CTrees don’t work well when classes are separable via linear equations
10Which of these is a feature of Gradiant Boosting?
- ATrees require you to assume linear relationships between X’s and Ys
- BIt is an average of multiple decision trees
- CIt is a weighted average of re-sampled data
- DIt could handle hundreds and thousands of input variables with no degeneracy in accuracy
11Which of these is not a strength of Decision Trees?
- ALess Appropriate for estimation tasks
- BCan handle all types of variables (categorical and continuous)
- CRequires little computing resource compared to other machine-based learning methods
- DIs very understandable for different business leaders
12Decision Trees are categorically classified as...
- ASupervised machine learning
- BUnsupervised learning
- CRegression based model
- DReinforcement learning algorithm
13Which of the following is true about Random Forest and Gradient Boosting ensemble methods?
- ABoth methods can be used for regression as well as classification tasks
- BRandom Forest is used for classification whereas Gradient Boosting is used for regression task
- CRandom Forest is used for regression whereas Gradient Boosting is used for Classification task
- DBoth methods can be used for neither regression nor classification tasks
14Which of the following algorithm is not an example of ensemble learning algorithm?
- ADecision Trees
- BGradient Boosting
- CExtra Trees
- DRandom Forest
15All of the following are features of gradient boosting except:
- Athe series of base learners combined together does not form a single predictive model
- Ba series of base learner models are created
- Cthe results is a weighted average of re sampled data
- Ddata set resampled several times
16All of the following is true about random forests except:
- Aa random is forest is not an average of multiple decision trees
- Brandom forest approach could handle hundreds and thousands of input variables with no degeneration in accuracy
- Cthe training data is a random sample of the original data.
- Da branch search is performed on a random set of inputs
Información
- Descripción
- Test sobre Module 4 Review, creado por Nick Feldman el 01/05/2022.
- Autor
- Nick Feldman
- Idioma
- English
- Visitas
- 0
- Valoración
- Aún sin valoraciones Valora este test