sadimanna

def specificity(y, pred, th=0.5):
    tn = TN(y,pred,th)
    fp = FP(y,pred,th)
    return tn/(tn+fp)

sadimanna

def fscore(y,pred,beta=1.0):
    p = precision(y,pred)
    r = recall(y,pred)
    return (1+beta**2)*p*r/((1+beta**2)*p+r)

sadimanna

def npv(y, pred, th=0.5):
    tn = TN(y,pred,th)
    fn = FN(y,pred,th)
    return tn/(tn+fn)

sadimanna

def prevalence(y):
    return np.sum(y)/y.shape[0]

sadimanna

def prevalence(y):
    return np.sum(y)/y.shape[0]

sadimanna

def fdr(y,pred,th=0.5):
    fp = FP(y,pred,th)
    tp = TP(y,pred,th)
    return fp/(fp+tp)

sadimanna

def for(y,pred,th=0.5):
    fn = FN(y,pred,th)
    tn = TN(y,pred,th)
    return fn/(fn+tn)

sadimanna

def fnr(y,pred,th=0.5):
    fn = FN(y,pred,th)
    tp = TP(y,pred,th)
    return fn/(tp+fn)

sadimanna

def fpr(y,pred,th=0.5):
    fp = FP(y,pred,th)
    tn = TN(y,pred,th)
    return fp/(tn+fp)

sadimanna

def roc_curve(y,pred):
    TPR = [0.0]
    FPR = [0.0]
    thresholds = np.arange(0.01,1.00+0.01,0.01)
    for th in thresholds:
        TPR.append(recall(y,pred,th))
        FPR.append(fpr(y,pred,th))
        
    return TPR,FPR,thresholds