NormalizeScore function

Normalize Score using Max and Min normalization

ReduceAnomalies It reduces the number of detected anomalies. This function is designed to reduce the number of false positives keeping only the first detection of all those that are close to each other. This proximity distance is defined by a window

NormalizeScore(real.score, perfect.score, null.score)

Arguments

• real.score: Detector score. See GetDetectorScore.
• perfect.score: Perfect detector score; one that outputs all true positives and no false positives. See GetNullAndPerfectScores.
• null.score: Perfect detector score; one that outputs all true positives and no false positives. See GetNullAndPerfectScores.

Returns

Normalized score.

Examples

## Generate data
set.seed(100)
n <- 180
x <- sample(1:100, n, replace = TRUE)
x[70:90] <- sample(110:115, 21, replace = TRUE)
x[25] <- 200
x[150] <- 170
df <- data.frame(timestamp = 1:n, value = x)

# Add is.real.anomaly column
df$is.real.anomaly <- 0
df[c(25,80,150), "is.real.anomaly"] <- 1

## Calculate anomalies
result <- CpSdEwma(
  data = df$value,
  n.train = 5,
  threshold = 0.01,
  l = 3
)
res <- cbind(df, result)

# Get null and perfect scores
np.scores <- GetNullAndPerfectScores(df)
np.standard <- np.scores[1,]
np.fp <- np.scores[2,]
np.fn <- np.scores[3,]

# Get detector score
scores <- GetDetectorScore(res, print = FALSE, title = "")

# Normalize standard score
NormalizeScore(scores$standard, np.standard$perfect.score, np.standard$null.score)

# Normalize low_FP_rate score
NormalizeScore(scores$low_FP_rate, np.fp$perfect.score, np.fp$null.score)

# Normalize low_FN_rate score
NormalizeScore(scores$low_FN_rate, np.fn$perfect.score, np.fn$null.score)

References

A. Lavin and S. Ahmad, “Evaluating Real-time Anomaly Detection Algorithms – the Numenta Anomaly Benchmark,” in 14th International Conference on Machine Learning and Applications (IEEE ICMLA’15), 2015.