How should I practice Dog Retrieval Precision and Recall?

Start by reading the full statement, solve the visible examples, then compare the pattern against related Microsoft onsite interview problems in FastPrep.

Is Dog Retrieval Precision and Recall associated with Microsoft?

FastPrep tags this as a Microsoft onsite interview coding practice problem when the problem catalog has that company and stage context.

Dog Retrieval Precision and Recall

Microsoft● EasyNEW GRADONSITE INTERVIEW

Learn

Problem statement

A corpus contains exactly ten image files, identified by 0 through 9. The entry labels[i] is 1 when file i depicts a dog and 0 for another animal. A retrieval API was asked for k files, and its received response is represented by returnedIds and responsePresent.

Return [precision, recall] using these rules:

If responsePresent is false, the response is None. Ignore returnedIds and return [-1.0, -1.0].
A present response containing a duplicate ID or an ID outside 0..9 is invalid. Return [-1.0, -1.0] for the whole response.
Otherwise, compute metrics from all actual returned files, even when their count differs from k. Precision is the number of returned dogs divided by the actual returned count. Recall is the number of returned dogs divided by the total number of dogs in the corpus.
Define a metric with denominator zero as 0.0. A present empty response is valid and differs from a missing response.

The received result is supplied as input; no image classifier or external API call is required. Absolute error of at most 10^-9 is accepted.

Function

dogRetrievalMetrics(labels: int[], returnedIds: int[], k: int, responsePresent: boolean) → double[]

Examples

Example 1

labels = [1,0,1,0,0,1,0,0,0,1]returnedIds = [0,1,5]k = 3responsePresent = truereturn = [0.6666666666666666,0.5]

Two of the three returned files are dogs, and the corpus contains four dogs.

Example 2

labels = [1,0,1,0,0,1,0,0,0,1]returnedIds = [0,1]k = 5responsePresent = truereturn = [0.5,0.25]

The API returned only two files despite requesting five. Precision uses the actual two results, while recall still uses all four corpus dogs.

Example 3

labels = [1,0,1,0,0,1,0,0,0,1]returnedIds = []k = 4responsePresent = falsereturn = [-1.0,-1.0]

The API response is None, represented by responsePresent=false. Metrics are unavailable rather than zero.

Constraints

labels.length = 10, and each label is 0 or 1.
0 <= k <= 10.
0 <= returnedIds.length <= 100; each received ID is between -100 and 100.
responsePresent is a boolean.