Decision tree diagram

Decision trees in three modes — taxonomy (yes/no question flows), decision analysis (expected value with probabilities), and machine learning (trained classifier visualization). Used by product managers, business analysts, data scientists, clinicians, and customer support teams.

decisiontree·§ —

↘ preview

100%

Line 4: Orphan line (bad indent): yes: question "Team size > 50?"

UTF-8 · LF · 9 lines · 286 chars✗ parse error

Three modes in one keyword: taxonomy flows, probability-weighted decision analysis, and ML classifier trees.

Modes

decisiontree           # taxonomy (default) — yes/no question trees
decisiontree:decision  # decision analysis — expected value + probabilities
decisiontree:ml        # machine learning — trained classifier visualization

Taxonomy mode (default)

Best for: troubleshooting guides, FAQs, clinical decision support, product recommendation flows.

decisiontree "Title"

question "Root question?"
  yes: question "Follow-up?"
    yes: answer "Outcome A"
    no: answer "Outcome B"
  no: answer "Outcome C"

Node keywords: question (or q), answer (or a, leaf)

Branch labels: yes:, no:, or custom — label "Low risk":

Config options

direction: left-right   # left-right (default) | top-down
branchLabels: boolean   # boolean (default) | relation
edgeStyle: diagonal     # diagonal | orthogonal | bracket

Custom branch labels

decisiontree "Triage"

question "Pain level?"
  label "Severe": answer "Emergency — go to ER"
  label "Moderate": answer "Urgent care"
  label "Mild": answer "Rest + OTC medication"

Decision analysis mode

Best for: investment decisions, risk analysis, strategy evaluation, financial modeling with expected values.

decisiontree:decision "Build vs Buy"

decision "Build in-house?"
  choice "Build"
    chance "Succeeds" [payoff: 500000]
      prob 0.7
        end "High ROI" [payoff: 800000]
      prob 0.3
        end "Over budget" [payoff: 100000]
  choice "Buy vendor"
    end "Known cost" [payoff: 400000]

Node keywords: decision, chance, end, outcome

Branch keywords: choice "label", prob 0.6

Properties: [payoff: value] on any node

Expected value is calculated automatically via backward induction.

Constraint

Probabilities on chance node children must sum to 1.0 (±0.01).

Machine learning mode

Best for: explaining trained classifiers, feature importance analysis, model transparency reports.

decisiontree:ml "Iris Classification"
direction: top-down
impurity: gini

split "Petal Length <= 2.45" feature=petal_length op="<=" threshold=2.45 samples=150 gini=0.666
  true
    leaf "Setosa" value=50 gini=0
  false
    split "Petal Width <= 1.75" feature=petal_width op="<=" threshold=1.75 samples=100 gini=0.5
      true
        leaf "Versicolor" value=50 gini=0
      false
        leaf "Virginica" value=50 gini=0

Node keywords: split, leaf

Branch prefixes: true, false

ML properties (key=value, no colon):

Property	Meaning
`feature=name`	Feature used at split
`op="<="`	Comparison operator
`threshold=5.9`	Split threshold
`samples=150`	Sample count
`gini=0.5`	Gini impurity
`entropy=0.5`	Entropy
`class=name`	Class for leaf
`value=50`	Sample count or `[50,50]` array

Config options (ML)

direction: top-down   # top-down | left-right
impurity: gini        # gini | entropy | mse | gain
classes: A, B, C      # class labels for display

Examples

Troubleshooting guide

decisiontree "Email Not Sending"

question "Error shown in UI?"
  yes: question "Error code 550?"
    yes: answer "Recipient email rejected — check address"
    no: answer "Check SMTP logs for error detail"
  no: question "Message in Sent folder?"
    yes: answer "Delivery delayed — check spam filters"
    no: answer "Client not connected — check network"

Medical triage

decisiontree "Chest Pain Triage"
edgeStyle: orthogonal

question "Onset sudden?"
  yes: question "ECG changes?"
    yes: answer "ACS protocol — cardiology consult"
    no: question "D-dimer elevated?"
      yes: answer "PE workup — CT pulmonary angio"
      no: answer "Aortic dissection — CT angio"
  no: question "Reproducible on palpation?"
    yes: answer "Musculoskeletal — NSAIDs, follow-up"
    no: answer "GI / anxiety — further history"

Investment decision analysis

decisiontree:decision "Market Entry"
direction: top-down

decision "Enter new market?"
  choice "Enter"
    chance "Market reception"
      prob 0.5
        end "Strong adoption" [payoff: 2000000]
      prob 0.3
        end "Moderate adoption" [payoff: 800000]
      prob 0.2
        end "Poor adoption" [payoff: -300000]
  choice "Stay out"
    end "Status quo" [payoff: 0]

Decision tree diagram

On this page