شغل أصيل

Selected Answer: True
Correct Answer: True
Question 3
Machine Learning is great for:
D. complex problems for which we have no algorithmic solution.

Selected Answer:
B. All answers are correct.

Correct Answer:
Question 4
Consider the following dataset:
ID Height Age Target-case

1 5 45 sick
2 5.11 26 sick
3 5.6 30 Not sick
4 5.9 34 Not sick
5 4.8 40 sick
6 5.8 36 Not sick
7 5.3 19 Not sick
8 5.8 28 sick
Which of the following code lines can be used for the above dataset
Selected Answers:
from sklearn.neighbors import KNeighborsClassifier
C.
from sklearn.neighbors import KNeighborsRegressor

D.
Correct Answers:
from sklearn.linear_model import LogisticRegression
A.
from sklearn.neighbors import KNeighborsClassifier

C.
Question 5
Please put the following scikit-learn codes for the iris dataset in the correct order.
Correct Answer Selected Answer
from sklearn.datasets import load_iris from sklearn.datasets import load_iris

iris = load_iris() iris = load_iris()
1. 1.
X = iris["data"] X = iris["data"]
y = iris.target y = iris.target
2. 2.
3. 3.
from sklearn.model_selection import train_test_split from sklearn.model_selection import train_te

X_train,X_test, y_train,y_test = train_test_split(X,y, test_size=0.2, random_state=4) X_train,X_test, y_train,y_test = train_test_
from sklearn.neighbors import KNeighborsClassifier from sklearn.neighbors import KNeighbor

knn = KNeighborsClassifier(n_neighbors=5) knn = KNeighborsClassifier(n_neighbors=
4. 4.
knn.fit(X_train, y_train) knn.fit(X_train, y_train)

5. 5.
y_pred=knn.predict(X_test) y_pred=knn.predict(X_test)
6. 6.
from sklearn import metrics from sklearn import metrics

score = metrics.accuracy_score(y_test,y_pred) score = metrics.accuracy_score(y_test,y_
7. 7.
Question 6
Gradient Descent cannot get stuck in a local minimum when training a Logistic Regression model because the cost
function is convex
Selected Answer: False

Question 7
Suppose you are using Polynomial Regression. You plot the learning curves and you notice that there is a large gap
between the training error and the validation error.
The issue is that the model is overfitting the training set. This can be solved by:
Selected D.
Answer: Regularizing the model. For example, by adding an l2 penalty (Ridge) or an l1 penalty (Lasso) to the cost
function.
B. All answers are correct

Correct
Answer:
Question 8
For the good ML practice, it is recommended to split the dataset into three sets:
______________ set, Validation set & Testing set
Selected Answer: Training set

Correct Answer:
Evaluation Method Correct Answer Case Sensitivity
Contains Training
Contains training
Contains Train
Contains train
Question 9
model = SGDRegressor(max_iter=1000, tol=1e-3, penalty="l1", eta0=0.1)
l1 = Lasso
The training approach that is applied for the above code in scikit-learn tool is:
B. Stochastic Gradient Descent with indication penalty for regularization.

Selected Answer:
B. Stochastic Gradient Descent with indication penalty for regularization.

Correct Answer:
Question 10
Logistic =[A] is a =[B] supervised learning algorithm that is used to estimate the =[C] that an instance belongs to a
particular =[D].
Specified Answer for: A regression
Specified Answer for: B classification
Specified Answer for: C probability
Specified Answer for: D class
Correct Answers for: A
Exact Match regression
Exact Match Regression
Correct Answers for: B
Exact Match classification
Exact Match Classification
Correct Answers for: C
Exact Match probability
Exact Match Probability
Correct Answers for: D
Exact Match class
Exact Match Class
Exact Match type
Exact Match Type
Exact Match group
Exact Match Group
Question 11
"It is the science (and art) of programming computers so they can learn from data."
B. Machine Learning
Selected Answer:
B. Machine Learning
Correct Answer:
Question 12
How many parameters are in a linear regression model for the dataset with 5 features?
C. 7
Selected Answer:
B. 6
Correct Answer:
Question 13
The reason for the above performance of the linear regression model is:
A. The convex shape of MSE cost function

Selected Answer:
C. The very small value of the learning rate

Correct Answer:
Question 14
Suppose you want to classify pictures as outdoor/indoor and daytime/nighttime.
Which of the following classifiers you should implement?
D. Two Logistic Regression classifiers.

Selected Answer:
D. Two Logistic Regression classifiers.

Correct Answer:
Question 15
Examples x1 x2 class
1 6 1 0
2 7 3 0
3 8 2 0
4 9 0 0
5 8 4 1
6 8 6 1
7 9 2 1
8 9 5 1
To classify a new example (8, 3.5) using kNN classifier with manhattan distance, then the correct
answers to the following blanks are:
The closest example from the dataset is x1 = [A] & x2 = [B]
The distance to the closest example is = [C]
The class of the new example is = [D]
Specified Answer for: A 4
Specified Answer for: B 4
Specified Answer for: C 4
Specified Answer for: D 1
Exact Match 8
Exact Match 4
Exact Match 0.5
Exact Match .5
Exact Match 1/2
Exact Match 1
Question 16
In order to regularize a linear regression model, we should add:
Selected Answer:
B. to the cost function.

Correct Answer:
B. to the cost function.

Question 17
To predict the prices of cars, the type of Machine Learning algorithm that would you use is:
C. Supervised Regression Learning

Selected Answer:
C. Supervised Regression Learning

Correct Answer:
Question 18
Fill in the blank by selecting the correct answer:
___________ is a generalization approach using a similarity measure to compare new examples to

learned examples.
D. Instance-based Learning
Selected Answer:
D. Instance-based Learning
Correct Answer:
Question 19
Let's assume the following partial code:
model = LogisticRegression()
model.fit(input_train, target_train)
If you have been asked to find the output of the sigmoid function, then the correct line code using
scikit-learn is: sigmoid function ‫داﻣﮫ ﻗﺎل‬
Selected Answer:
‫ﻣﻌﻨﺎﺗﮫ ﻳ�� اﺣﺘﻤﺎﻟﻴﺔ‬
output_value = model.predict_proba(input_test)
B.
Correct Answer:
B.
Question 20
Select the appropriate answers that describe the kNN model.
A. It is an instance-based learning approach.

Selected Answers:
D. Non-parametric learning algorithm
A. It is an instance-based learning approach.

Correct Answers:
D. Non-parametric learning algorithm
Question 21
Consider a simple dataset:
Example Feature 1 Feature 2 Target

1 1 -2 Red
2 15 -5 Green
3 20 1 Green
4 5 3 Green
5 3 30 Red
6 23 19 Red
7 50 22 Green
If the values of Feature 1 are replaced by:

-0.9245221
-0.100857
0.19330917
-0.6891892
-0.8068556
0.36980884
1.95830589
This process should be called:
(Hint)
mean of feature_1= 16.714
standard deviation of feature_1= 16.997

1 ‫�� ﻗﻴﻤﺔ أﻛ�� ﻣﻦ‬
B. Normalization
Selected Answer:
D. Standardization
Correct Answer:
Saturday, June 11, 2022 10:49:07 PM AST
← OK
6/11/22, 10:47 PM Review Test Submission: Midterm (April 19 2022) – ‫ مقدمة‬...
D. Confusion Matrix
Question 2
"It is the study of how to make computers do things at which, at the moment, people are better."
D. Artificial Intelligence
Selected Answer:
D. Artificial Intelligence
Correct Answer:
Question 3
Consider a simple dataset:
Example Feature 1 Feature 2 Target

1 1 -2 Red
2 15 -5 Green
3 20 1 Green
4 5 3 Green
5 3 30 Red
6 23 19 Red
7 50 22 Green
If the values of Feature 1 are replaced by:

-0.9245221
-0.100857
0.19330917
-0.6891892
-0.8068556
0.36980884
1.95830589
This process should be called:
(Hint)
D. Normalization
Selected Answer:
A. Standardization
Correct Answer:
Question 4
model = SGDRegressor(max_iter=1000, tol=1e-3, penalty="l1", eta0=0.1)
https://lms.uqu.edu.sa/webapps/assessment/review/review.jsp?attempt_id=_7983773_1&course_id=_220895_1&content_id=_1938118_1&outcome_i… 2/7
The training approach that is applied for the above code in scikit-learn tool is:
C. Stochastic Gradient Descent with indication penalty for regularization.

Selected Answer:
C. Stochastic Gradient Descent with indication penalty for regularization.

Correct Answer:
Question 5
Model parameters should be the optimal values in order to use the model for predicting new
instances.

Question 6
Please put the following scikit-learn codes for the iris dataset in the correct order.
Correct Answer Selected Answer
from sklearn.datasets import load_iris from sklearn.datasets import load_iris

iris = load_iris() iris = load_iris()
1. 1.
X = iris["data"] X = iris["data"]
y = iris.target y = iris.target
2. 2.
3. 3.
from sklearn.model_selection import train_test_split from sklearn.model_selection import train_te

X_train,X_test, y_train,y_test = train_test_split(X,y, test_size=0.2, random_state=4) X_train,X_test, y_train,y_test = train_test_
from sklearn.neighbors import KNeighborsClassifier from sklearn.neighbors import KNeighbors

knn = KNeighborsClassifier(n_neighbors=5) knn = KNeighborsClassifier(n_neighbors=5
4. 4.
knn.fit(X_train, y_train) knn.fit(X_train, y_train)

5. 5.
y_pred=knn.predict(X_test) y_pred=knn.predict(X_test)
6. 6.
from sklearn import metrics from sklearn import metrics

score = metrics.accuracy_score(y_test,y_pred) score = metrics.accuracy_score(y_test,y_
7. 7.
Question 7
Gradient Descent cannot get stuck in a local minimum when training a Logistic Regression model because the cost
function is convex

Question 8
For the good ML practice, it is recommended to split the dataset into three sets:
______________ set, Validation set & Testing set
Selected Answer: Training set

Correct Answer:
Contains Training
Contains training
Contains Train
Contains train
Question 9
Select the appropriate answers that describe the kNN model.
C. Only for classification tasks

Selected Answers:
D. It is an instance-based learning approach.
B. Non-parametric learning algorithm

Correct Answers:
D. It is an instance-based learning approach.
Question 10
ID Height Age Target-Weight

1 5 45 77
2 5.11 26 47
3 5.6 30 55
4 5.9 34 59
5 4.8 40 72
6 5.8 36 60
7 5.3 19 40
8 5.8 28 45
Which of the following code lines can be used for the above dataset
Selected Answers:
B.
Correct Answers:
from sklearn.neighbors import KNeighborsRegressor
A.
from sklearn.linear_model import LinearRegression

C.
Question 11
Let's assume the following partial code:
model = LogisticRegression()
model.fit(input_train, target_train)
If you have been asked to find the class type, then the correct line code using scikit-learn is:
Selected Answer:
B.
Correct Answer:
output_value = model.predict(input_test)
C.
Question 12
Machine Learning is great for:
C. All answers are correct.

Selected Answer:
C. All answers are correct.

Correct Answer:
Question 13
Fill in the blank by selecting the correct answer
___________ is a learning approach that is used for training a model using all the available data.
A. Model-based Learning
Selected Answer:
‫ ﻛﻤﻴﺔ اﻟﺪاﺗﺎ اﳌﺴﺘﺨﺪﻣﺔ‬، ‫اﻟﻈﺎهﺮ ان اﳌﻘﺼﻮد‬
‫ اﻧﮫ ﻛﻴﻒ �ﺴﺘﺨﺪﻣهﺎ‬... ‫وﻟ�ﺲ‬
B. Batch Learning
Correct Answer:
Question 14
To detect spam filters, the type of Machine Learning algorithm that would you use is:
D. Supervised Classification Learning

Selected Answer:
D. Supervised Classification Learning

Correct Answer:
Question 15
Suppose you want to classify pictures as outdoor/indoor and daytime/nighttime.
Which of the following classifiers you should implement?
A. Two Logistic Regression classifiers.

Selected Answer:
A. Two Logistic Regression classifiers.

Correct Answer:
Question 16
The reason for the above performance of the linear regression model is:
A. The very large value of the learning rate

Selected Answer:
D. The very small value of the learning rate

Correct Answer:
Question 17
In order to regularize a linear regression model, we should add:
C. Non Of the answers

Selected Answer:
Correct Answer:
A. to the cost function.
Question 18
Logistic =[A] is a =[B] supervised learning algorithm that is used to estimate the =[C] that an instance belongs to a
particular =[D].
Specified Answer for: A regression
Specified Answer for: B classification
Specified Answer for: C propability
Specified Answer for: D class
Exact Match regression
Exact Match Regression
Exact Match classification
Exact Match Classification
Exact Match probability
Exact Match Probability
Exact Match class
Exact Match Class
Exact Match type
Exact Match Type
Exact Match group
Exact Match Group
Question 19
How many parameters are in a linear regression model for the dataset with 6 features?
D. 6
Selected Answer:
A. 7
Correct Answer:
Question 20
Suppose you are using Polynomial Regression. You plot the learning curves and you notice that there is a large gap
between the training error and the validation error.
The issue is that the model is overfitting the training set. This can be solved by:
Selected Answer:
C. Increasing the size of the training set.
A. All answers are correct

Correct Answer:
Question 21
Examples x1 x2 class
1 6 1 0
2 7 3 0
3 8 2 0
4 9 0 0
5 8 4 1
6 8 6 1
7 9 2 1
8 9 5 1
To classify a new example (8, 3.5) using kNN classifier with manhattan distance, then the correct
answers to the following blanks are:
The closest example from the dataset is x1 = [A] & x2 = [B]
The distance to the closest example is = [C]
The class of the new example is = [D]
Specified Answer for: A 1
Specified Answer for: B 6
Specified Answer for: C 4
Specified Answer for: D 1
Exact Match 8
Exact Match 4
Exact Match 0.5
Exact Match .5
Exact Match 1/2
Exact Match 1
Saturday, June 11, 2022 10:46:45 PM AST
← OK

شغل أصيل

Uploaded by

Copyright:

Available Formats

شغل أصيل

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

شغل أصيل

Uploaded by

Copyright:

Available Formats

Selected Answer: True

Correct Answer: True

Machine Learning is great for:

D. complex problems for which we have no algorithmic solution.

B. All answers are correct.

Consider the following dataset:

ID Height Age Target-case

from sklearn.neighbors import KNeighborsRegressor

from sklearn.neighbors import KNeighborsClassifier

Correct Answer Selected Answer

from sklearn.datasets import load_iris from sklearn.datasets import load_iris

from sklearn.model_selection import train_test_split from sklearn.model_selection import train_te

from sklearn.neighbors import KNeighborsClassifier from sklearn.neighbors import KNeighbor

knn.fit(X_train, y_train) knn.fit(X_train, y_train)

from sklearn import metrics from sklearn import metrics

Selected Answer: False

B. All answers are correct

______________ set, Validation set & Testing set

Selected Answer: Training set

model = SGDRegressor(max_iter=1000, tol=1e-3, penalty="l1", eta0=0.1)

B. Stochastic Gradient Descent with indication penalty for regularization.

B. Stochastic Gradient Descent with indication penalty for regularization.

Specified Answer for: A regression

Specified Answer for: B classification

Specified Answer for: C probability

Specified Answer for: D class

Correct Answers for: A

Evaluation Method Correct Answer Case Sensitivity

Exact Match regression

Exact Match Regression

Correct Answers for: B

Evaluation Method Correct Answer Case Sensitivity

Exact Match classification

Exact Match Classification

Correct Answers for: C

Evaluation Method Correct Answer Case Sensitivity

Exact Match probability

Exact Match Probability

Correct Answers for: D

Evaluation Method Correct Answer Case Sensitivity

Exact Match class

Exact Match Class

Exact Match type

Exact Match Type

Exact Match group

Exact Match Group

A. The convex shape of MSE cost function

C. The very small value of the learning rate

Suppose you want to classify pictures as outdoor/indoor and daytime/nighttime.

Which of the following classifiers you should implement?

D. Two Logistic Regression classifiers.

D. Two Logistic Regression classifiers.

Consider the following dataset:

The closest example from the dataset is x1 = [A] & x2 = [B]

The distance to the closest example is = [C]

The class of the new example is = [D]

Specified Answer for: A 4

Specified Answer for: B 4

Specified Answer for: C 4

Specified Answer for: D 1