Data Mining - Decision Tree Induction

decision_trees.rb

def entropy(total, yes, no)
	yes_div = (yes.to_f/total.to_f)
	no_div = (no.to_f/total.to_f)
	yes_log = (yes == 0 ) ? 0 : -((yes_div) * (Math.log2(yes_div)))
	no_log = (no == 0) ? 0 : ((no_div) * (Math.log2(no_div)))
	yes_log - no_log
end

def info(total, args, gain)
	value = 0
	args.each do |key, val|
		value += (((key.to_f/total) * val.to_f))
	end

	(value - gain).abs
end

gain = 0.940
humidity_high = entropy(7, 3, 4)
humidity_normal = entropy(7, 6, 1)
humidity_hash = [
	[7, humidity_high],
	[7, humidity_normal]	
]
puts "Humidity \t\t#{info(14, humidity_hash, gain)}"

temperature_hot = entropy(4, 2, 2)
temperature_mild = entropy(6, 4, 2)
temperature_cold = entropy(4, 2, 2)
temperature_hash = [
	[4, temperature_hot],
	[6, temperature_mild],
	[4, temperature_cold]
]
puts "Temperature \t\t#{info(14,  temperature_hash, gain)}"

windy_true = entropy(8, 6, 2)
windy_false = entropy(6, 3, 3)
windy_hash = [
	[8, windy_true], 
	[6, windy_false]
]
puts "Windy \t\t#{info(14, windy_hash, gain)}"

puts "SUNNY"
sunny_temperature_hot = entropy(2, 0, 2)
sunny_temperature_mild = entropy(2, 1, 1)
sunny_temperature_cold = entropy(1, 1, 0)
sunny_temperature_hash = [
	[2, sunny_temperature_hot],
	[2, sunny_temperature_mild],
	[1, sunny_temperature_cold]
]
gain = 0.97
puts "Gain Temperature #{info(5,  sunny_temperature_hash, gain)}"

sunny_windy_true = entropy(2, 1, 1)
sunny_windy_false = entropy(3, 1, 2)
sunny_windy_hash = [
	[2, sunny_windy_true], 
	[3, sunny_windy_false]
]
gain = 0.97
puts "Gain Windy #{info(5, sunny_windy_hash, gain)}"

sunny_humidity_high = entropy(3, 0, 3)
sunny_humidity_normal = entropy(2, 2, 0)
sunny_humidity_hash = [
	[3, sunny_humidity_high], 
	[2, sunny_humidity_normal]
]
gain = 0.97
puts "Gain Humidity #{info(5, sunny_humidity_hash, gain)}"

puts "RAINY"
rainy_temperature_hot = entropy(0, 0, 0)
rainy_temperature_mild = entropy(3, 2, 1)
rainy_temperature_cold = entropy(2, 1, 1)
rainy_temperature_hash = [
	[0, rainy_temperature_hot],
	[3, rainy_temperature_mild],
	[2, rainy_temperature_cold]
]
gain = 0.97
puts "Gain Temperature #{info(5,  rainy_temperature_hash, gain)}"

rainy_windy_true = entropy(2, 0, 2)
rainy_windy_false = entropy(3, 3, 0)
rainy_windy_hash = [
	[2, rainy_windy_true], 
	[3, rainy_windy_false]
]
gain = 0.97
puts "Gain Windy #{info(5, rainy_windy_hash, gain)}"

rainy_humidity_high = entropy(2, 1, 1)
rainy_humidity_normal = entropy(3, 2, 1)
rainy_humidity_hash = [
	[3, rainy_humidity_high], 
	[2, rainy_humidity_normal]
]
gain = 0.97
puts "Gain Humidity #{info(5, rainy_humidity_hash, gain)}"