R/init_prob.R
sens.Rdsens defines a decision's sensitivity (or hit rate) value:
The conditional probability of the decision being positive
if the condition is TRUE.
sensAn object of class numeric of length 1.
Understanding or obtaining the sensitivity sens
(or hit rate HR):
Definition: sens is the conditional probability
for a (correct) positive decision given that
the condition is TRUE:
sens = p(decision = positive | condition = TRUE)
or the probability of correctly detecting true cases
(condition = TRUE).
Perspective:
sens further classifies
the subset of cond_true individuals
by decision (sens = hi/cond_true).
Alternative names:
true positive rate (TPR),
hit rate (HR),
probability of detection,
power = 1 - beta,
recall
Relationships:
a. sens is the complement of the miss rate
mirt (aka. false negative rate FNR or the
rate of Type-II errors):
sens = (1 - miss rate) = (1 - FNR)
b. sens is the opposite conditional probability
-- but not the complement --
of the positive predictive value PPV:
PPV = p(condition = TRUE | decision = positive)
In terms of frequencies,
sens is the ratio of
hi divided by
cond_true (i.e., hi + mi):
sens = hi/cond_true = hi/(hi + mi)
Dependencies:
sens is a feature of a decision process
or diagnostic procedure and a measure of
correct decisions (true positives).
Due to being a conditional probability,
the value of sens is not intrinsic to
the decision process, but also depends on the
condition's prevalence value prev.
Consult Wikipedia for additional information.
comp_sens computes sens as the complement of mirt;
prob contains current probability information;
comp_prob computes current probability information;
num contains basic numeric parameters;
init_num initializes basic numeric parameters;
comp_freq computes current frequency information;
is_prob verifies probabilities.
Other probabilities:
FDR,
FOR,
NPV,
PPV,
acc,
err,
fart,
mirt,
ppod,
prev,
spec
sens <- .85 # sets a sensitivity value of 85%
sens <- 85/100 # (decision = positive) for 85 out of 100 people with (condition = TRUE)
is_prob(sens) # TRUE
#> [1] TRUE