Modified Z-Score by Claude.ai

main.go

package main

import (
	"fmt"
	"math"
	"sort"
)

// RemoveOutliersSmallDataset removes outliers from a small dataset using the Modified Z-score method
// threshold is typically 3.5 (more conservative) to 3.0 (more aggressive)
func RemoveOutliersSmallDataset(prices []float64, threshold float64) []float64 {
	if len(prices) <= 3 {
		return prices // Not enough data to determine outliers
	}

	// Calculate the median
	sortedCopy := make([]float64, len(prices))
	copy(sortedCopy, prices)
	sort.Float64s(sortedCopy)

	var median float64
	n := len(sortedCopy)
	if n%2 == 0 {
		median = (sortedCopy[n/2-1] + sortedCopy[n/2]) / 2
	} else {
		median = sortedCopy[n/2]
	}

	// Calculate the Median Absolute Deviation (MAD)
	deviations := make([]float64, len(prices))
	for i, price := range prices {
		deviations[i] = math.Abs(price - median)
	}

	sort.Float64s(deviations)
	var mad float64
	if n%2 == 0 {
		mad = (deviations[n/2-1] + deviations[n/2]) / 2
	} else {
		mad = deviations[n/2]
	}

	// Constant factor for normal distribution
	const c = 0.6745

	// Handle the case where MAD is zero (can happen with repeated values)
	if mad == 0 {
		// Alternative approach when MAD is zero - use a small epsilon or
		// simply return the original data since there's no variation
		return prices
	}

	// Filter based on Modified Z-scores
	result := []float64{}
	for _, price := range prices {
		modifiedZScore := math.Abs((price - median) / (mad * c))
		if modifiedZScore <= threshold {
			result = append(result, price)
		}
	}

	return result
}

// Usage example
func main() {
	prices := []float64{10.2, 14.1, 14.4, 14.4, 14.4, 14.5, 14.5, 14.6, 14.7, 15.0, 45.0}
	filteredPrices := RemoveOutliersSmallDataset(prices, 3.5)

	fmt.Println("Unfiltered", prices)
	// Unfiltered [10.2 14.1 14.4 14.4 14.4 14.5 14.5 14.6 14.7 15 45]
	fmt.Println("Filtered", filteredPrices)	
	// Filtered [14.4 14.4 14.4 14.5 14.5 14.6 14.7]
 
}