Description
Compare expected binary prediction costs.
A classification decision can compare expected losses or utilities under supplied probabilities and outcome costs. The resulting threshold depends on which outcomes are charged, the cost signs and whether abstention is allowed; arithmetic alone does not establish appropriate costs or reliable probabilities.
Compare expected binary prediction costs.
When to use this tool
Use Binary Prediction Flip Cost Term when comparing the stated classification, rejection or utility terms under supplied assumptions.
How to use the tool and how it works
- Enter falsePositiveCost, falseNegativeCost, positiveProbability. Use the units and conventions below.
- Run the tool. Correct any format, range, or cross-field input error before interpreting the output.
- Read positiveCost, negativeCost, preferPositive. Inspect preferPositive before interpreting the associated values.
Compare expected binary prediction costs. Require probability in [0,1] and nonnegative FP/FN costs; positiveCost = FPcost*(1-p), negativeCost = FNcost*p, preferPositive uses strict lower positive cost.1
| Input field | Meaning and units | Accepted input |
|---|---|---|
| falsePositiveCost | False Positive Cost | Required. Finite number. |
| falseNegativeCost | False Negative Cost | Required. Finite number. |
| positiveProbability | Positive Probability | Required. Finite number; minimum 0; maximum 1. |
| Output field | Interpretation | Returned shape |
|---|---|---|
| positiveCost | Positive Cost | Required. Finite number. |
| negativeCost | Negative Cost | Required. Finite number. |
| preferPositive | Strictly lower positiveCost; false includes exact ties. | Required. Boolean. |
{
"input": {
"falsePositiveCost": 2,
"falseNegativeCost": 6,
"positiveProbability": 0.5
},
"output": {
"positiveCost": 1,
"negativeCost": 3,
"preferPositive": true
}
}Limitations and assumptions
- An exact cost tie returns preferPositive false rather than a distinct tie selection; actual costs and posterior validity are not assessed.
- Inputs must satisfy both the field types and the cross-field conditions above. Invalid input causes an error; a successful false flag or documented placeholder is a different result.
- Numeric calculations use binary floating point unless exact decimal-integer arithmetic is explicitly stated. No hidden measurement tolerance or decimal rounding is added; nonfinite or out-of-contract results are rejected. Numeric signed-zero results are exported as zero.
Alternative or Complementary methods
Specify the full decision problem and validate probabilities and outcome costs before using a threshold; assess performance and consequences on relevant data.
References
Similar or alternative tools
- Binary Rejection Cost Indifference
Return expected positive, negative and reject risks.
- Abstention Cost Indifference Threshold
Return correct-probability indifference between classification and abstention.