hasMinimumPrecision(_:forRecall:)
Determines whether the observation has a minimum precision value for a specific recall.
Declaration
func hasMinimumPrecision(_ minimumPrecision: Float, forRecall recall: Float) -> BoolParameters
- minimumPrecision:
The minimum desired percentage of correct positive classifications.
- recall:
The percentage of all positive classifications that the algorithm correctly classified.
Return Value
A Boolean value that indicates whether the classification observation provides a minimum percentage of correct results that meet the desired recall.
Discussion
The following example uses the hasMinimumPrecision(_:forRecall) method to perform a high-recall filter on the results of a ClassifyImageRequest:
let results = try await request.perform(on: image)
.filter { $0.hasMinimumPrecision(0.1, forRecall: 0.8) }A high-recall filter retains a much broader range of observations, but can result in more false positive results. Testing can help determine the balance point between the minimumPrecision and recall values to return the best results for a specific use case.