Skip to content

Instantly share code, notes, and snippets.

@intrip
Created July 23, 2020 09:12
Show Gist options
  • Save intrip/f6760ce2a4dcdd7dbc7a73f304f7fc13 to your computer and use it in GitHub Desktop.
Save intrip/f6760ce2a4dcdd7dbc7a73f304f7fc13 to your computer and use it in GitHub Desktop.
require 'matrix'
require 'ruby_linear_regression'
ROWS_PER_OPTION = 1_000
POSITION_VALUES_MAPPING = {
top_left: { value: [1, 1], consent_rate: 60 },
bottom_left: { value: [1000, 1], consent_rate: 30 },
top_right: { value: [1, 1000], consent_rate: 45 },
bottom_right: { value: [1000, 1000], consent_rate: 20 },
}
generated_array = []
POSITION_VALUES_MAPPING.each do |position, values_map|
consent_rate = values_map[:consent_rate]
lower_bound = consent_rate - 0.1 * consent_rate
upper_bound = consent_rate + 0.1 * consent_rate
ROWS_PER_OPTION.times do
random_consent_rate = rand(lower_bound..upper_bound)
generated_array << [values_map[:value], random_consent_rate]
end
end
generated_array.shuffle!
x_data = []
y_data = []
POSITION_MAPPING = {
top_left: [1, 1],
bottom_left: [1000, 1],
top_right: [1, 1000],
bottom_right: [1000, 1000],
}
# Load data from CSV file into two arrays - one for independent variables X (x_data) and one for the dependent variable y (y_data)
generated_array.each do |row|
# Each row contains option value and expected consent rate
x_data.push( [row[0][0].to_i, row[0][1].to_i])
y_data.push( row[1].to_i )
end
# Create regression model
linear_regression = RubyLinearRegression.new
# Load training data
linear_regression.load_training_data(x_data, y_data)
# Train the model using gradient descent
# p linear_regression.train_gradient_descent(0.0001, 1000, false)
# Train using least squares
linear_regression.train_normal_equation
# Output the cost
puts "Trained model with the following cost fit #{linear_regression.compute_cost}"
# Predictions
POSITION_MAPPING.each do |key, value|
predicted_consent_rate = linear_regression.predict([value[0], value[1]])
puts "Prediction for #{key} position is #{predicted_consent_rate}"
end
#Trained model with the following cost fit 3.6977304062500846
#
# Prediction for top_left position is 58.2812499999997
# Prediction for bottom_left position is 30.821750000000435
# Prediction for top_right position is 45.663750000000164
# Prediction for bottom_right position is 18.204250000000897
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment