Classification Tools
Browse all available classification tools.
Available tools
- Abstention Cost Indifference Threshold Tool Solve the supplied expected-cost equality p*Ccorrect+(1−p)*Cwrong=Cabstain without selecting a decision policy.
- Abstention Zero Coverage Policy Tool Explicit zero-coverage reporting policy; undefined accuracy uses a flagged zero placeholder.
- Accuracy Calculator Calculate classification accuracy from true and false outcome counts.
- Balanced Accuracy Calculator Calculate the mean of sensitivity and specificity.
- Balanced Accuracy Single Class Term Calculate one explicit class recall contribution to unweighted multiclass balanced accuracy.
- Binary Abstention Coverage Count Tool Calculate binary abstention coverage count from explicit caller-supplied inputs.
- Binary Abstention Indicator Tool Calculate binary abstention indicator from explicit caller-supplied inputs.
- Binary Balanced Error Contribution Tool Calculate binary balanced error contribution from explicit caller-supplied inputs.
- Binary Class Weight Effective Prevalence Tool Compute binary class weight effective prevalence from explicit bounded inputs.
- Binary Confusion Predicted Prevalence Term Tool Calculate binary confusion predicted prevalence term from explicit caller-supplied inputs.
- Binary Confusion Transpose Role Change Tool Calculate binary confusion transpose role change from explicit caller-supplied inputs.
- Binary Logit Temperature Slope Tool Evaluate sigmoid(logit/temperature) and its derivative with respect to the supplied logit.
- Binary Matthews Denominator Factor Tool Calculate binary matthews denominator factor from explicit caller-supplied inputs.
- Binary Predicted Label Count Balance Reconcile explicit binary confusion counts into predicted-label totals using exact integer arithmetic.
- Binary Prediction Flip Cost Term Tool Compute binary prediction flip cost term from explicit bounded inputs.
- Binary Prior Shift Logit Offset Tool Compute binary prior shift logit offset from explicit bounded inputs.
- Binary Rejection Cost Indifference Tool Compare action risks and exact indifference with a caller-supplied rejection cost.
- Binary Specificity Minimum Count Margin Tool Compute binary specificity minimum count margin from explicit bounded inputs.
- Brier Uncertainty Single Class Term Tool Calculate brier uncertainty single class term from explicit caller-supplied inputs.
- Class Label Cost Tie Selector Tool Select an exact minimum-cost label with an explicit stable tie policy.
- Class Probability Renormalization Abstain Term Tool Compute class probability renormalization abstain term from explicit bounded inputs.
- Class Recall Minimum Count Margin Tool Compute class recall minimum count margin from explicit bounded inputs.
- Class Weight Rebalancing Multiplier Tool Calculate class weight rebalancing multiplier from explicit caller-supplied inputs.
- Classification: Abstention Correct Denominator Tool Compute accuracy under an explicit include/exclude-abstention denominator, with attempted counts excluding abstentions and undefined status retained.
- Classification: Binary Confusion Positive Label Swap Tool Swap the designated positive and negative classes in one binary confusion record, preserving actual/predicted roles.
- Classification: Binary Label Smoothing Target Pair Tool Apply uniform-two-class label smoothing with explicit epsilon in [0,1], returning both target components rather than a predicted class.
- Classification: Binary Probability Rounding Tie Tool Round one binary probability to a class using an explicit half-probability tie policy; does not infer calibration.
- Classification: Decision Threshold To Cost Ratio Tool Derive the calibrated binary posterior decision threshold from positive false-positive/false-negative costs with zero correct-decision cost; not a fitted calibration or recommendation.
- Classification: Fraction To Class Weight Ratio Tool Derive the positive-to-negative inverse-frequency class-weight ratio from an explicit binary training fraction; not an inferred prevalence or fairness criterion.
- Classification: Label Prior Correction Odds Factor Tool Compute binary posterior-odds prior-shift correction under caller-established unchanged class-conditional distributions; no calibration or domain-shift validation.
- Classification: Ordinal Step Error Distance Tool Measure signed and absolute ordinal class-index error under an explicitly supplied ordered class count.
- Classification: Precision Zero Denominator Policy Tool Compute precision with an explicit zero/one undefined-denominator display convention while retaining a separate defined flag.
- Classifier Logit Tie Policy Tool Compute classifier logit tie policy from explicit bounded inputs.
- Classifier Probability Sum Roundoff Gap Tool Compute classifier probability sum roundoff gap from explicit bounded inputs.
- Classifier Threshold Equality Policy Tool Apply an explicit positive-class equality policy to a finite scalar classifier threshold.
- Classifier Threshold Open Closed Band Tool Compute classifier threshold open closed band from explicit bounded inputs.
- Classwise False Discovery Count Balance Tool Calculate classwise false discovery count balance from explicit caller-supplied inputs.
- Confusion Actual Support Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; report actual row support totals without calculating per-class rates.
- Confusion Cell Expected Independence Term Tool Compute confusion cell expected independence term from explicit bounded inputs.
- Confusion Cell Row Share Column Share Normalize one explicit confusion cell to supplied positive row and column totals without choosing an axis convention.
- Confusion Column Rate Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; normalize each predicted column independently with null for zero-support columns; no calibrated probability claim.
- Confusion Diagonal Correct Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; report diagonal correct counts separately by label and in total; not a normalized accuracy estimate.
- Confusion Error Flow Imbalance Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; compare incoming versus outgoing off-diagonal error weight for each label, retaining signed class error-flow imbalance.
- Confusion Group Table Sum Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; sum aligned explicit subgroup confusion observations by group, preserving one output table per exact group name; no hidden label-axis alignment.
- Confusion Joint Rate Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; normalize joint observed weight by total support, preserving null on a zero-total table; does not estimate a joint population distribution.
- Confusion Label Axis Reorder Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; reorder both axes by a complete explicit label permutation without changing counts or label identity.
- Confusion Label Class Collapse Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; collapse both axes through explicit per-label group names, summing every count exactly once; first group appearance defines output order.
- Confusion Largest Offdiagonal Pairs Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; return every positive maximum off-diagonal error pair, retaining ties; an error-free table has maximum zero and no invented error pair.
- Confusion One Versus Rest Records Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; derive full one-versus-rest TP/FP/FN/TN records for every label; no division, threshold choice or prevalence adjustment.
- Confusion Predicted Support Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; report predicted column support totals without calculating per-class rates.
- Confusion Reject Column Partition Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; partition one explicit predicted reject column without removing actual rows or relabeling rejected observations as errors.
- Confusion Row Diagonal Dominance Gap Tool Calculate confusion row diagonal dominance gap from explicit caller-supplied inputs.
- Confusion Row Rate Table Explicit nonnegative square confusion table with unique supplied labels, actual rows and predicted columns; normalize each actual row independently with null for all cells of zero-support rows; undefined evidence is not converted to zero.
- Confusion Table Transpose Convention Tool Normalize a square count confusion matrix into actual-row/predicted-column orientation using the caller's declared original convention.
- F1 Score Calculator Calculate the harmonic mean of classification precision and recall.
- Fbeta One Cell Sensitivity Tool Calculate fbeta one cell sensitivity from explicit caller-supplied inputs.
- Focal Loss One Probability Term Tool Calculate focal loss one probability term from explicit caller-supplied inputs.
- Hierarchical Label Depth Error Term Tool Compute hierarchical label depth error term from explicit bounded inputs.
- Hierarchical Label Parent Constraint Pair Tool Compute hierarchical label parent constraint pair from explicit bounded inputs.
- Label Pair Confusion Table Build an occurrence-count multiclass confusion table over explicit actual/predicted label observations with a caller-ordered complete label universe; unknown labels reject, no identity coercion.
- Label Pair Coverage Audit Audit unknown label occurrences separately in aligned actual and predicted observations against an explicit label universe; do not drop or coerce unknown evidence.
- Label Pair Disagreement Selector Select exact actual/predicted label disagreement observations with original indexes, preserving repeated labels; no classification threshold or error severity model.
- Multiclass Cross Entropy Smoothing Term Tool Calculate multiclass cross entropy smoothing term from explicit caller-supplied inputs.
- Multiclass Score Mass Deficit Tool Compare supplied nonnegative class masses with an explicit target without normalizing or asserting calibration.
- Multilabel Exact Match Single Observation Tool Compute multilabel exact match single observation from explicit bounded inputs.
- Multilabel Item Excess Prediction Count Return unique explicitly predicted labels absent from the actual-label set while preserving prediction order.
- Multilabel Subset Violation Witness Tool Compute multilabel subset violation witness from explicit bounded inputs.
- Ordinal Cumulative Label Target Tool Compute ordinal cumulative label target from explicit bounded inputs.
- Ordinal Label Adjacent Threshold Gap Subtract explicit adjacent ordinal cutpoints, retaining nonpositive gaps as ordering diagnostics.
- Ordinal Label Cost Matrix Cell Tool Compute ordinal label cost matrix cell from explicit bounded inputs.
- Ordinal Median Class Boundary Tool Select a supplied ordinal mass median with an explicit exact-half boundary convention.
- Ordinal Probability Mass Interval Tool Inclusive interval of supplied ordered class indices, with a declared 1e-12 normalization tolerance.
- Positive Label Threshold Tie Tool Calculate positive label threshold tie from explicit caller-supplied inputs.
- Precision Calculator Calculate positive predictive value from true and false positive counts.
- Precision Recall Breakeven Residual Tool Calculate precision recall breakeven residual from explicit caller-supplied inputs.
- Predicted Class Margin Runner Up Tool Calculate predicted class margin runner up from explicit caller-supplied inputs.
- Predicted Prior Mass Residual Tool Compare supplied candidate class masses without normalizing them or asserting calibration.
- Prediction Explicit Threshold Reject Map Apply a supplied inclusive maximum-score threshold to explicit unit-sum probability rows, rejecting unresolved top ties rather than guessing a label; not a validated decision policy.
- Prediction Reject Band Membership Tool Compute prediction reject band membership from explicit bounded inputs.
- Prediction Set Cardinality Penalty Tool Calculate prediction set cardinality penalty from explicit caller-supplied inputs.
- Prediction Tie Random Mass Share Tool Compute prediction tie random mass share from explicit bounded inputs.
- Prediction Top Label Tie Audit Audit every maximum-probability label tie from explicit nonnegative unit-sum score rows; retain all ties and do not assume calibration or pick a hidden winner.
- Recall Calculator Calculate sensitivity from true positive and false negative counts.
- ROC AUC Calculator Area under the ROC curve via the trapezoidal rule over descending scores; equal scores share one threshold point so ties are handled exactly.
- Selective Classification Accepted Count Subtract explicit abstentions from prediction counts and report coverage, using a flagged zero sentinel for an empty total.
- Selective Risk From Accepted Errors Tool Calculate selective risk from accepted errors from explicit caller-supplied inputs.
- Specificity Calculator Calculate true negative rate from true negative and false positive counts.
- Threshold Utility Indifference Probability Tool Calculate threshold utility indifference probability from explicit caller-supplied inputs.
- Top K Boundary Membership Flag Compare an explicit score with a supplied kth-score boundary under explicit score direction and tie policies.
- Top K Tie Slot Budget Tool Calculate top k tie slot budget from explicit caller-supplied inputs.
- Top Two Score Relative Gap Tool Normalize the two greatest supplied scores’ separation by absolute top score, preserving signed-score inputs.
- Weighted Label Pair Confusion Table Build a weighted multiclass confusion table from aligned actual/predicted observations and nonnegative supplied weights; zero weights retained as evidence, no probability inference.