Machine Learning Basics — Free Practice Questions

Q: Accurate labels are necessary to properly train supervised models. Many times, human subjects known as raters perform this labeling effort. What are the main categories of human raters? (check all that apply).

Correct answer: Subject matter experts; Your users. Generalists usually come from crowdsourcing sites. SME or subject matter experts : classical example is radiologists labeling medical images for automated diagnosis tools. Users can provide labels within your application. A classical example is photo tagging. Generalists usually come from crowdsourcing sites. SME or subject matter experts : classical example is radiologists labeling medical images for automated diagnosis tools. Users can provide labels within your application. A classical example is photo tagging.

Q: A cardiologist labeling MRI images is a typical example of Direct Labeling.

Correct answer: False. MRI labeling by a medical specialist is an example of Human Labeling and not Direct Labeling.

Q: Distribution skew occurs when the distribution of the training dataset is significantly different from the distribution of the serving dataset, and is typically caused by: (check all that apply).

Correct answer: Different data sources for training and serving data.; Trend, seasonality, changes in data over time.; Faulty sampling method that selects a sample for training which is not representative of serving data distribution.. Distribution Skew example : 1) A faulty sampling mechanism that chooses a non-representative subsample is an example of distribution skew. 2) Data sources between training and serving often change and so this is another case of distribution skew. 3) Data distributions between training and serving often change and so this is another case of distribution skew. Distribution Skew example : 1) A faulty sampling mechanism that chooses a non-representative subsample is an example of distribution skew. 2) Data sources between training and serving often change and so this is another case of distribution skew. 3) Data distributions between training and serving often change and so this is another case of distribution skew. Distribution Skew example : 1) A faulty sampling mechanism that chooses a non-representative subsample is an example of distribution skew. 2) Data sources between training and serving often change and so this is another case of distribution s

Q: TensorFlow Data Validation (TFDV) helps TFX users maintain the health of their ML pipelines. TFDV can analyze training and serves data to:

Correct answer: Detect data anomalies.; Infer a schema.; Compute descriptive statistics.. TensorFlow Data Validation (TFDV) performs Descriptive Statistics , Infer a schema and Anomaly Detection. It does not perform Feature Engineering or Selection. Deployment on mobile device is done by TensorFlow Lite. TensorFlow Data Validation (TFDV) performs Descriptive Statistics , Infer a schema and Anomaly Detection. It does not perform Feature Engineering or Selection. Deployment on mobile device is done by TensorFlow Lite. TensorFlow Data Validation (TFDV) performs Descriptive Statistics , Infer a schema and Anomaly Detection. It does not perform Feature Engineering or Selection. Deployment on mobile device is done by TensorFlow Lite.

Q: Sigmoid function is

Correct answer: Continuous; Differential; Smooth. Sigmoid function produces an output between 0 and 1 and not -1 to 1 Sigmoid function produces an output between 0 and 1 and not -1 to 1 Sigmoid function produces an output between 0 and 1 and not -1 to 1

Q: If model training time is of essence, what criteria would you choose for a Decision Tree Classifier ?

Correct answer: Gini. The answer lies in the mathematics behind the two criteria. Computationally, entropy is more complex since it makes use of logarithms and consequently, the calculation of the Gini Index will be faster.

Q: Which of the following are examples of text preprocessing?

Correct answer: Lowercasing; Stemming; Removing stopwords, punctuation, handles and URLs. Stemming, or the process of reducing a word to its word stem and Lowercasing, which is the process of changing all capital letter to lower case are text processing techniques. Also, we remove stop words, URL, punctuation and handles. Stemming, or the process of reducing a word to its word stem and Lowercasing, which is the process of changing all capital letter to lower case are text processing techniques. Also, we remove stop words, URL, punctuation and handles. Stemming, or the process of reducing a word to its word stem and Lowercasing, which is the process of changing all capital letter to lower case are text processing techniques. Also, we remove stop words, URL, punctuation and handles.

Q: Which of the following might be good ways to help prevent a data leakage situation?

Correct answer: Remove variables that a model in production wouldn’t have access to; If time is a factor, remove any data related to the event of interest that doesn’t take place prior to the event.; Sanity check the model with an unseen validation set. All options are correct except 'Ensure that data is preprocessed outside of any cross validation folds' Data leakage is a big problem in machine learning when developing predictive models. Data leakage is when information from outside the training dataset is used to create the model. Note : if any other feature whose value would not actually be available in practice at the time you’d want to use the model to make a prediction, is a feature that can introduce leakage to your model All options are correct except 'Ensure that data is preprocessed outside of any cross validation folds' Data leakage is a big problem in machine learning when developing predictive models. Data leakage is when information from outside the training dataset is used to create the model. Note : if any other feature whose value would not actually be available in practice at the time you’d want to use the model to make a prediction, is a feature that can introduc

Question 1

1. Accurate labels are necessary to properly train supervised models. Many times, human subjects known as raters perform this labeling effort. What are the main categories of human raters? (check all that apply).

✓Subject matter experts
Generalists usually come from crowdsourcing sites. SME or subject matter experts : classical example is radiologists labeling medical images for automated diagnosis tools. Users can provide labels within your application. A classical example is photo tagging.
✓Your users
✗Aggregators
✗Generalists

Question 2

2. A cardiologist labeling MRI images is a typical example of Direct Labeling.

✓False
MRI labeling by a medical specialist is an example of Human Labeling and not Direct Labeling.
✗True

Question 3

3. Distribution skew occurs when the distribution of the training dataset is significantly different from the distribution of the serving dataset, and is typically caused by: (check all that apply).

✓Different data sources for training and serving data.
Distribution Skew example : 1) A faulty sampling mechanism that chooses a non-representative subsample is an example of distribution skew. 2) Data sources between training and serving often change and so this is another case of distribution skew. 3) Data distributions between training and serving often change and so this is another case of distribution skew.
✓Trend, seasonality, changes in data over time.
✓Faulty sampling method that selects a sample for training which is not representative of serving data distribution.
✗There is different logic for generating features between training and serving. For example, if you apply some transformation only in one of the two code paths.

Question 4

4. TensorFlow Data Validation (TFDV) helps TFX users maintain the health of their ML pipelines. TFDV can analyze training and serves data to:

✓Detect data anomalies.
TensorFlow Data Validation (TFDV) performs Descriptive Statistics , Infer a schema and Anomaly Detection. It does not perform Feature Engineering or Selection. Deployment on mobile device is done by TensorFlow Lite.
✓Infer a schema.
✓Compute descriptive statistics.
✗Deploy pipeline to a mobile application.

Question 5

5. Sigmoid function is

✗ranges from -1 to 1
✓Continuous
Sigmoid function produces an output between 0 and 1 and not -1 to 1
✓Differential
✓Smooth

Question 6

6. If model training time is of essence, what criteria would you choose for a Decision Tree Classifier ?

✓Gini
The answer lies in the mathematics behind the two criteria. Computationally, entropy is more complex since it makes use of logarithms and consequently, the calculation of the Gini Index will be faster.
✗Entropy

Question 7

7. Which of the following are examples of text preprocessing?

✓Lowercasing
Stemming, or the process of reducing a word to its word stem and Lowercasing, which is the process of changing all capital letter to lower case are text processing techniques. Also, we remove stop words, URL, punctuation and handles.
✓Stemming
✗Adding new words to make sure all the sentences make sense
✓Removing stopwords, punctuation, handles and URLs

Question 8

8. Which of the following might be good ways to help prevent a data leakage situation?

✗Ensure that data is preprocessed outside of any cross validation folds
✓Remove variables that a model in production wouldn’t have access to
All options are correct except 'Ensure that data is preprocessed outside of any cross validation folds' Data leakage is a big problem in machine learning when developing predictive models. Data leakage is when information from outside the training dataset is used to create the model. Note : if any other feature whose value would not actually be available in practice at the time you’d want to use the model to make a prediction, is a feature that can introduce leakage to your model
✓If time is a factor, remove any data related to the event of interest that doesn’t take place prior to the event.
✓Sanity check the model with an unseen validation set

Machine Learning Basics — Free Practice Questions

Topics covered

Sample questions with answers