time_series_fit.rb

require 'gsl'
require 'time'

module Diskalerter
  class ThresholdEstimator
    attr_reader :timestamps, :signal, :threshold, :now

    # timestamps - a list of timestamps in epoch milliseconds
    # signal - a list of values for the provided timestamps
    # threshold - the "exhaustion" threshold (e.g., 100 for a percentage signal)
    def initialize(timestamps, signal, threshold)
      @timestamps = timestamps
      @signal = signal
      @threshold = threshold
      @now = Time.now.to_i
    end

    def alertable?
      now < intersect && intersect < warning_threshold
    end

    def will_intersect?
      now < intersect && intersect < never
    end

    def intersect
      x_intersect, slope = linear_fit
      (threshold - x_intersect) / slope
    end

    def linear_fit
      return @linear_fit if @linear_fit

      timestamps, signal = filtered_data

      x = GSL::Vector[*timestamps]
      y = GSL::Vector[*signal]

      @linear_fit = GSL::Fit.linear(x, y)
    end

    def warning_threshold
      now + one_week
    end

    def one_week
      3600 * 24 * 7
    end

    def five_years
      one_week * 52 * 5
    end

    def never
      now + five_years
    end

    def intersect_time
      Time.at(intersect)
    end

    def intersect_days
      (intersect - now) / 24 / 3600
    end

    def intersect_milliseconds
      (intersect - now) * 1000
    end

    def intersect_time_string
      if will_intersect?
        '%s (%0.2f days)' % [intersect_time.iso8601, intersect_days]
      else
        'Never'
      end
    end

    private

    def filtered_data
      timestamps.map { |t| t / 1000 }.zip(signal).reject { |pair| pair[1].nil? }.transpose
    end
  end
end