Class: DecisionAgent::Simulation::MonteCarloSimulator

Inherits:
Object
  • Object
show all
Defined in:
lib/decision_agent/simulation/monte_carlo_simulator.rb

Overview

Monte Carlo simulator for probabilistic decision outcomes

Allows you to model input variables with probability distributions and run simulations to understand decision outcome probabilities.

Examples:

simulator = MonteCarloSimulator.new(agent: agent)

# Define probabilistic inputs
distributions = {
  credit_score: { type: :normal, mean: 650, stddev: 50 },
  amount: { type: :uniform, min: 50_000, max: 200_000 }
}

# Run simulation
results = simulator.simulate(
  distributions: distributions,
  iterations: 10_000,
  base_context: { name: "John Doe" }
)

puts "Decision probabilities: #{results[:decision_probabilities]}"
puts "Average confidence: #{results[:average_confidence]}"

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(agent:, version_manager: nil) ⇒ MonteCarloSimulator

Returns a new instance of MonteCarloSimulator.



33
34
35
36
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 33

def initialize(agent:, version_manager: nil)
  @agent = agent
  @version_manager = version_manager || Versioning::VersionManager.new
end

Instance Attribute Details

#agent ⇒ Object (readonly)

Returns the value of attribute agent.



31
32
33
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 31

def agent
  @agent
end

#version_manager ⇒ Object (readonly)

Returns the value of attribute version_manager.



31
32
33
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 31

def version_manager
  @version_manager
end

Instance Method Details

#analyze_field_sensitivity(base_distributions, field, param_variations, iterations, base_context, options) ⇒ Object 



133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 133

def analyze_field_sensitivity(base_distributions, field, param_variations, iterations, base_context, options)
  param_variations.each_with_object({}) do |(param_name, param_values), field_results|
    config = {
      base_distributions: base_distributions,
      field: field,
      param_name: param_name,
      param_values: param_values,
      iterations: iterations,
      base_context: base_context,
      options: options
    }
    param_results = run_parameter_variations(config)
    field_results[param_name] = build_parameter_result(param_name, param_values, param_results)
  end
end

#analyze_sensitivity_params(base_distributions, sensitivity_params, iterations, base_context, options) ⇒ Object 



125
126
127
128
129
130
131
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 125

def analyze_sensitivity_params(base_distributions, sensitivity_params, iterations, base_context, options)
  sensitivity_params.each_with_object({}) do |(field, param_variations), results|
    results[field] = analyze_field_sensitivity(
      base_distributions, field, param_variations, iterations, base_context, options
    )
  end
end

#build_param_result(param_value, result) ⇒ Object 



171
172
173
174
175
176
177
178
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 171

def build_param_result(param_value, result)
  {
    param_value: param_value,
    decision_probabilities: result[:decision_probabilities],
    average_confidence: result[:average_confidence],
    confidence_intervals: result[:confidence_intervals]
  }
end

#build_parameter_result(param_name, param_values, param_results) ⇒ Object 



180
181
182
183
184
185
186
187
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 180

def build_parameter_result(param_name, param_values, param_results)
  {
    parameter: param_name,
    values_tested: param_values,
    results: param_results,
    impact_analysis: analyze_parameter_impact(param_results)
  }
end

#create_modified_distribution(base_distributions, field, param_name, param_value) ⇒ Object 



164
165
166
167
168
169
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 164

def create_modified_distribution(base_distributions, field, param_name, param_value)
  modified = base_distributions.dup
  modified[field] = modified[field].dup
  modified[field][param_name] = param_value
  modified
end

#run_parameter_variations(config) ⇒ Object 



149
150
151
152
153
154
155
156
157
158
159
160
161
162
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 149

def run_parameter_variations(config)
  config[:param_values].map do |param_value|
    modified_distributions = create_modified_distribution(
      config[:base_distributions], config[:field], config[:param_name], param_value
    )
    result = simulate(
      distributions: modified_distributions,
      iterations: config[:iterations],
      base_context: config[:base_context],
      options: config[:options].merge(parallel: false)
    )
    build_param_result(param_value, result)
  end
end

#sensitivity_analysis(base_distributions:, sensitivity_params:, iterations: 5_000, base_context: {}, options: {}) ⇒ Hash

Run sensitivity analysis using Monte Carlo simulation Varies one distribution parameter at a time to see its impact

Parameters:

  • base_distributions (Hash) —

    Base probabilistic input distributions

  • sensitivity_params (Hash) —

    Hash of field => parameter variations Example: { credit_score: { mean: [600, 650, 700], stddev: [40, 50, 60] } }

  • iterations (Integer) (defaults to: 5_000) —

    Number of iterations per sensitivity test

  • base_context (Hash) (defaults to: {}) —

    Base context values

  • options (Hash) (defaults to: {}) —

    Simulation options

Returns:

  • (Hash) —

    Sensitivity analysis results



98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 98

def sensitivity_analysis(
  base_distributions:,
  sensitivity_params:,
  iterations: 5_000,
  base_context: {},
  options: {}
)
  options = {
    parallel: true,
    thread_count: 4,
    seed: nil,
    confidence_level: 0.95
  }.merge(options)

  srand(options[:seed]) if options[:seed]

  sensitivity_results = analyze_sensitivity_params(
    base_distributions, sensitivity_params, iterations, base_context, options
  )

  {
    sensitivity_results: sensitivity_results,
    base_distributions: base_distributions,
    iterations_per_test: iterations
  }
end

#simulate(distributions:, iterations: 10_000, base_context: {}, rule_version: nil, options: {}) ⇒ Hash

Run Monte Carlo simulation with probabilistic input distributions

Parameters:

  • distributions (Hash) —

    Hash of field_name => distribution_config Distribution configs support:

    • { type: :normal, mean: Float, stddev: Float } - Normal distribution
    • { type: :uniform, min: Numeric, max: Numeric } - Uniform distribution
    • { type: :lognormal, mean: Float, stddev: Float } - Log-normal distribution
    • { type: :exponential, lambda: Float } - Exponential distribution
    • { type: :discrete, values: Array, probabilities: Array } - Discrete distribution
    • { type: :triangular, min: Numeric, mode: Numeric, max: Numeric } - Triangular distribution
  • iterations (Integer) (defaults to: 10_000) —

    Number of Monte Carlo iterations (default: 10_000)

  • base_context (Hash) (defaults to: {}) —

    Base context values that are fixed (not probabilistic)

  • rule_version (String, Integer, Hash, nil) (defaults to: nil) —

    Optional rule version to use

  • options (Hash) (defaults to: {}) —

    Simulation options

    • :parallel [Boolean] Use parallel execution (default: true)
    • :thread_count [Integer] Number of threads (default: 4)
    • :seed [Integer] Random seed for reproducibility (default: nil)
    • :confidence_level [Float] Confidence level for intervals (default: 0.95)

Returns:

  • (Hash) —

    Simulation results with decision probabilities and statistics



57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
 
# File 'lib/decision_agent/simulation/monte_carlo_simulator.rb', line 57

def simulate(distributions:, iterations: 10_000, base_context: {}, rule_version: nil, options: {})
  options = {
    parallel: true,
    thread_count: 4,
    seed: nil,
    confidence_level: 0.95
  }.merge(options)

  # Set random seed for reproducibility
  srand(options[:seed]) if options[:seed]

  # Validate distributions
  validate_distributions!(distributions)

  # Build agent from version if specified
  analysis_agent = build_agent_from_version(rule_version) if rule_version
  analysis_agent ||= @agent

  # Run Monte Carlo iterations
  results = run_iterations(
    distributions: distributions,
    base_context: base_context,
    iterations: iterations,
    agent: analysis_agent,
    options: options
  )

  # Calculate statistics (pass requested iterations count)
  calculate_statistics(results, options[:confidence_level], requested_iterations: iterations)
end