def estimateBias()

in src/main/scala/com/amazon/deequ/analyzers/catalyst/StatefulHyperloglogPlus.scala [259:297]

• 26 lines of code
• 7 McCabe index (conditional complexity)

  def estimateBias(e: Double): Double = {
    val estimates = RAW_ESTIMATE_DATA(P - 4)
    val numEstimates = estimates.length

    // The estimates are sorted so we can use a binary search to find the index of the
    // interpolation estimate closest to the current estimate.
    val nearestEstimateIndex = java.util.Arrays.binarySearch(estimates, 0, numEstimates, e) match {
      case ix if ix < 0 => -(ix + 1)
      case ix => ix
    }

    // Use square of the difference between the current estimate and the estimate at the given
    // index as distance metric.
    def distance(i: Int): Double = {
      val diff = e - estimates(i)
      diff * diff
    }

    // Keep moving bounds as long as the (exclusive) high bound is closer to the estimate than
    // the lower (inclusive) bound.
    var low = math.max(nearestEstimateIndex - K + 1, 0)
    var high = math.min(low + K, numEstimates)
    while (high < numEstimates && distance(high) < distance(low)) {
      low += 1
      high += 1
    }

    // Calculate the sum of the biases in low-high interval.
    val biases = BIAS_DATA(P - 4)
    var i = low
    var biasSum = 0.0
    while (i < high) {
      biasSum += biases(i)
      i += 1
    }

    // Calculate the bias.
    biasSum / (high - low)
  }