Test
Classification, clustering and feature learning
Este test tiene tipos de pregunta que se abren en el reproductor de siempre.
Preguntas del test (23)
1A model which requires a training dataset to 'train' classifiers is a
- Asupervised model
- Bunsupervised model
2A model which does not require a training dataset to 'train' classifiers and does not need pre-defined outcomes is a:
- Asupervised model
- Bunsupervised model
3To relate categorical response variables to a predictor variable we need to use:
- Aclustering
- Bclassification
- Cfeature learning
4Random Forest Classification also offers a means of embedded feature selection through [blank_start]mean decrease accuracy[blank_end] and [blank_start]mean decrease Gini[blank_end].
- Amean decrease accuracy
- Bmean decrease Gini
5In mean decrease accuracy, the [blank_start]greater[blank_end] the decrease in accuracy, the more important the variable is to the model you are fitting.
- Agreater
6Gini impurity is a way of measuring [blank_start]heterogeneity[blank_end].
- Aheterogeneity
7In Gini impurity, what score would reflect complete homogeneity.
- A0
- B1
8If a variable is associated with a big decline in Gini score, it means this feature is:
- Apotentially more important
- Bpotentially less important
9Support Vector Machines, logistic regression and Random Forest are all examples of [blank_start]classification[blank_end].
- Aclassification
10The hyperplane has:
- Aone more dimension than the axes
- Bone less dimension than the axes
11To see if the classification is any good, we can use a [blank_start]confusion matrix[blank_end].
- Aconfusion matrix
12true positives / ( true positives + false negatives ) =
- Asensitivity, how many are actually positive
- Bspecificity, how many are actually negative
13true negative / ( true negative + false positives ) =
- Asensitivity, how many are actually positive
- Bspecificity, how many are actually negative
14Which of the following is a worthless AUC (area under the curve) score?
- A0
- B0.1
- C0.5
- D1
15What is meant by an AUC (area under the curve) score?
- Athe closer to 1, the better the classification
- Bthe closer to 0, the better the classification
16If we know that there are groups, but not sure how many, we should use:
- Aclustering
- Bclassification
- Cfeature learning
17Clustering is
- Aan unsupervised model
- Ba supervised model
18For identifying groups to be clustered we can use:
- AK-means clustering
- Bhierarchical clustering
- Cfeature learning
- DRandom Forest regression
- Elogistic regression
- Flogistic clustering
- GElbow method
- HBIC / AIC
19To be able to tell how many clusters we should use, we can use the:
- AElbow method
- BK means clustering
- Chierarchical clustering
- DBIC / AIC
20In the BIC / AIC method of clustering, the optimum model is the one with the [blank_start]lowest[blank_end] BIC / AIC score
- Alowest
21If we know the categories already, but want to see if the data clusters the same without it, we can use:
- AAdjusted Rand Index
- BK means clustering
- Chierarchical clustering
- Delbow method
- EBIC / AIC
22Principal Component Analysis is an example of
- Aclustering
- Bclassification
- Cfeature learning
23PCs explain variation in the data and are uncorrelated to each other.
- AVerdadero
- BFalso
Información
- Descripción
- MIBTP Statistics Test sobre Classification, clustering and feature learning, creado por Bethany Richmond el 04/11/2018.
- Autor
- Bethany Richmond
- Idioma
- English
- Visitas
- 2
- Etiquetas
- Valoración
- Aún sin valoraciones Valora este test