Total points 6
1.
Question 1
Fill in the blanks with the correct answer according to the descriptions in the boxes below: 


1 / 1 point

1. Data mining. 2. Dimensionality reduction.


1. Dimensionality reduction. 2. Data Science.


> 1. Data mining. 2. Data Science.


1. Data Science. 2. Data mining.

Correct
That’s right! The “before” and “now” of performance and resource requirements are represented respectively by the Data Mining and Data Science concepts.

2.
Question 2
What does the X value represent? 



1 / 1 point

The cursed number of dimensions.


The worst number of features for making predictions.


> The optimal number of features.


The number of features that reaches the maximum classification error.

Correct
Exactly! The x-axis coordinate of this critical point represents the number of features required by the classifier to work at its best. 

3.
Question 3
Which of the following are problems of high dimensionality in model performance? (Select all that apply)

1 / 1 point

> Higher runtimes and system requirements

Correct
Correct! The more dimensions, the higher the system requirements. Therefore, dimensionality reduction helps optimize the system's performance.


Smaller hypothesis space.


> Solutions take longer to reach global optimum

Correct
Right on track! Very often, reaching a global optimum is a more difficult task when dealing with high-dimensional problems.


> The possibility of more correlated features is greater.

Correct
You’ve got it! When having more dimensions, it is possible to have more correlated features making the selection of the most relevant features a more difficult task.

4.
Question 4
What does the following line of code refer to?  count_params(model_n.trainable_variables)

1 / 1 point

> The number of training parameters for Model n.


The number of testing parameters for Model n.


The number of classes for Model n. 


The number of dimensions for Model n.

Correct
That’s right! This code line allows to count the number of training parameters for the input model.

5.
Question 5
The amount of training data available, the complexity of the decision surface, and the classifier type define the number of ____________ to be used

1 / 1 point

Spaces


Datasets


Models


> Features

Correct
That’s right! These three aspects define the amount of features that will be used in a machine learning problem. 

6.
Question 6
True Or False: Classification subspaces allow to minimize separation among classes, while regression subspaces are used for maximizing correlation between projected data and response variable.

1 / 1 point

True


> False

Correct
That’s right! Classification subspaces maximize the separation among classes, while regression intends to maximize the correlation between two variables.