Exam DP-100 topic 3 question 133 discussion

Actual exam question from Microsoft's DP-100

Question #: 133
Topic #: 3

HOTSPOT -

You create an Azure Machine Learning workspace and a dataset. The dataset includes age values for a large group of diabetes patients. You use the dp_mean function from the SmartNoise library to calculate the mean of the age value. You store the value in a variable named age_mean.

You must output the value of the interval rage of released mean values that will be returned 95 percent of the time.

You need to complete the code.

Which code values should you use? To answer, select the appropriate options in the answer area.

NOTE: Each correct selection is worth one point.

Show Suggested Answer

Suggested Answer:

by vish9 at May 13, 2023, 8:45 p.m.

Comments

Submit Cancel

jl420

7 months ago

Correct: get_accuracy(): This function is typically used to obtain the accuracy level of the computed mean based on a given confidence level. In this case, you want the 95% confidence interval, so you pass 0.95 to the function. privacy_usage_to_accuracy() and compute_privacy_usage() are not typically used for confidence interval calculations; instead, they are related to privacy budget calculations.

upvoted 2 times

...

phdykd

1 year, 10 months ago

# Assuming you have the privacy usage stored in a variable called 'usage' alpha = 0.05 # 1 - 0.95 (to get 95% confidence interval) https://github.com/opendp/smartnoise-samples/blob/master/analysis/accuracy_pitfalls.ipynb accuracy = age_mean.get_accuracy(alpha, usage) print(accuracy)

upvoted 3 times

...

snegnik

2 years ago

I think the ansver is get_accuracy(0.05) # get DP mean of age dp_D_tilde_mean_ages.append(sn.dp_mean( data = D_tilde, privacy_usage = {'epsilon': 1})) accuracy = dp_D_tilde_mean_ages[0].get_accuracy(0.05) https://github.com/opendp/smartnoise-samples/blob/master/analysis/accuracy_pitfalls.ipynb

upvoted 1 times

...

snegnik

2 years ago

I think the ansver is wrong, I cant find any info about privacy_usage_to_accuracy in documentation.

upvoted 1 times

...

fqc

2 years ago

From Bard: The compute_privacy_usage() method takes two parameters: epsilon: The epsilon value for differential privacy confidence: The confidence level The compute_privacy_usage() method returns the privacy usage for the given epsilon and confidence level. The privacy usage is a measure of how much noise has been added to the data to protect privacy. print(age_mean.compute_privacy_usage(0.05, 0.95)

upvoted 1 times

...

vish9

2 years ago

Not sure.

upvoted 2 times

...