Synthesis Feasibility Assessment

Thermodynamic Selectivity vs Success Rate

engine correlation_matrix

config (1 key)

{
  "variables": [
    "thermodynamic_selectivity",
    "synthesis_success_rate",
    "reaction_driving_force",
    "synthesis_temperature",
    "metathesis_reaction_feasibility"
  ]
}

expected results (2 keys)

{
  "reaction_driving_force↔synthesis_temperature": {
    "direction": "negative",
    "note": "Stronger driving force may enable lower synthesis temperatures"
  },
  "thermodynamic_selectivity↔synthesis_success_rate": {
    "direction": "positive",
    "note": "Higher selectivity should predict higher success"
  }
}

Logistic Regression: Thermodynamics → Synthesis Outcome

engine logistic_regression

config (4 keys)

{
  "classification": true,
  "predictors": [
    "reaction_driving_force",
    "thermodynamic_selectivity"
  ],
  "target": "synthesis_success_rate",
  "threshold": 0.5
}

expected results (2 keys)

{
  "accuracy": {
    "min": 0.7
  },
  "thermodynamic_selectivity_coefficient": {
    "direction": "positive"
  }
}

Random Forest: Full Feature Synthesis Prediction

engine random_forest

config (3 keys)

{
  "classification": true,
  "predictors": [
    "reaction_driving_force",
    "thermodynamic_selectivity",
    "precursor_selection_score",
    "synthesis_temperature",
    "thermodynamic_control_threshold"
  ],
  "target": "synthesis_success_rate"
}

expected results (2 keys)

{
  "accuracy": {
    "min": 0.78
  },
  "feature_importance_rank_1": {
    "expected": "thermodynamic_selectivity"
  }
}

Synthesis Assessment Engine Comparison

engine synth_assess_selectivity←logistic_synthesis_outcome, rf_synthesis_full

config (2 keys)

{
  "compare_to": [
    "logistic_synthesis_outcome",
    "rf_synthesis_full"
  ],
  "mode": "selectivity_analysis"
}

expected results (1 key)

{
  "selectivity_threshold": {
    "note": "Engine implements Bartel's selectivity metric directly"
  }
}