Safe Haskell	None
Language	Haskell2010

DataFrame.DecisionTree

Contents

Estimators
Classifier configuration
Regressor configuration
Fitted tree structure
Solver configuration (fills TreeConfig.linearSolverConfig)

Description

Interpretable decision trees on DataFrames (CART refined by TAO). The curated public surface: classifier/regressor configs, fitted-model records, and their Fit/Predict instances.

Synopsis

module DataFrame.DecisionTree.Model
data TreeConfig = TreeConfig {
- maxTreeDepth :: Int
- minSamplesSplit :: Int
- minLeafSize :: Int
- percentiles :: [Int]
- expressionPairs :: Int
- synthConfig :: SynthConfig
- taoIterations :: Int
- taoConvergenceTol :: Double
- columnOrdering :: ColumnOrdering
- useLinearSolver :: Bool
- linearSolverConfig :: SolverConfig
- minCarePointsForLinear :: Int
- pureReplacementLinear :: Bool
}
defaultTreeConfig :: TreeConfig
data SynthConfig = SynthConfig {
- maxExprDepth :: Int
- boolExpansion :: Int
- disallowedCombinations :: [(Text, Text)]
- complexityPenalty :: Double
- enableStringOps :: Bool
- enableCrossCols :: Bool
- enableArithOps :: Bool
- maxCategoricalSubsetCardinality :: Int
- perColumnQuota :: Maybe Int
}
defaultSynthConfig :: SynthConfig
newtype ColumnOrdering = ColumnOrdering (Map SomeTypeRep OrdDict)
orderable :: (Columnable a, Ord a) => ColumnOrdering
defaultColumnOrdering :: ColumnOrdering
withOrdFrom :: Columnable a => ColumnOrdering -> (Ord a => r) -> Maybe r
data RegTreeConfig = RegTreeConfig {
- rtMaxDepth :: !Int
- rtMinSamplesSplit :: !Int
- rtMinLeafSize :: !Int
- rtMinImpurityDecrease :: !Double
}
defaultRegTreeConfig :: RegTreeConfig
data Tree a
- = Leaf !a
- | Branch !(Expr Bool) !(Tree a) !(Tree a)
treeDepth :: Tree a -> Int
data SolverConfig = SolverConfig {
- scL1Lambda :: !Double
- scL2Lambda :: !Double
- scMaxIter :: !Int
- scTol :: !Double
- scSampleWeights :: !(Maybe (Vector Double))
}
defaultSolverConfig :: SolverConfig

Estimators

Fitted-model records with their Fit/Predict instances and the estimator classes (via DataFrame.Model).

module DataFrame.DecisionTree.Model

Classifier configuration

data TreeConfig #

Constructors

TreeConfig
Fields maxTreeDepth :: Int minSamplesSplit :: Int minLeafSize :: Int percentiles :: [Int] expressionPairs :: Int synthConfig :: SynthConfig taoIterations :: Int taoConvergenceTol :: Double columnOrdering :: ColumnOrdering useLinearSolver :: Bool linearSolverConfig :: SolverConfig minCarePointsForLinear :: Int pureReplacementLinear :: Bool

Instances

Instances details

(Columnable a, Ord a) => Fit TreeConfig (Expr a) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

Associated Types

type ModelOf TreeConfig (Expr a)
Instance details Defined in DataFrame.DecisionTree.Model type ModelOf TreeConfig (Expr a) = DecisionTreeClassifier a
type FrameReq TreeConfig (Expr a)
Instance details Defined in DataFrame.DecisionTree.Model type FrameReq TreeConfig (Expr a) = 'AnyFrame

Methods

fit :: TreeConfig -> Expr a -> FrameFor (Expr a) -> FitResult (FrameFor (Expr a)) (ModelOf TreeConfig (Expr a)) Source #

type FrameReq TreeConfig (Expr a) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

type FrameReq TreeConfig (Expr a) = 'AnyFrame

type ModelOf TreeConfig (Expr a) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

type ModelOf TreeConfig (Expr a) = DecisionTreeClassifier a

defaultTreeConfig :: TreeConfig #

data SynthConfig #

Constructors

SynthConfig
Fields maxExprDepth :: Int boolExpansion :: Int disallowedCombinations :: [(Text, Text)] complexityPenalty :: Double enableStringOps :: Bool enableCrossCols :: Bool enableArithOps :: Bool maxCategoricalSubsetCardinality :: Int perColumnQuota :: Maybe Int

Instances

Instances details

Show SynthConfig
Instance details Defined in DataFrame.DecisionTree.Types Methods showsPrec :: Int -> SynthConfig -> ShowS # show :: SynthConfig -> String # showList :: [SynthConfig] -> ShowS #
Eq SynthConfig
Instance details Defined in DataFrame.DecisionTree.Types Methods (==) :: SynthConfig -> SynthConfig -> Bool # (/=) :: SynthConfig -> SynthConfig -> Bool #

defaultSynthConfig :: SynthConfig #

newtype ColumnOrdering #

Which column types support ordering for splits. Register a type with orderable and combine with <>.

Constructors

ColumnOrdering (Map SomeTypeRep OrdDict)

Instances

Instances details

Monoid ColumnOrdering
Instance details Defined in DataFrame.DecisionTree.Types Methods mempty :: ColumnOrdering # mappend :: ColumnOrdering -> ColumnOrdering -> ColumnOrdering # mconcat :: [ColumnOrdering] -> ColumnOrdering #
Semigroup ColumnOrdering
Instance details Defined in DataFrame.DecisionTree.Types Methods (<>) :: ColumnOrdering -> ColumnOrdering -> ColumnOrdering # sconcat :: NonEmpty ColumnOrdering -> ColumnOrdering # stimes :: Integral b => b -> ColumnOrdering -> ColumnOrdering #

orderable :: (Columnable a, Ord a) => ColumnOrdering #

Register a type as orderable for decision-tree splits.

defaultColumnOrdering :: ColumnOrdering #

All standard numeric, text, and primitive types.

withOrdFrom :: Columnable a => ColumnOrdering -> (Ord a => r) -> Maybe r #

Run k with the Ord a instance recovered from the ordering registry, or Nothing when a is not registered.

Regressor configuration

data RegTreeConfig Source #

Stopping criteria for the regression tree.

Constructors

RegTreeConfig
Fields rtMaxDepth :: !Int rtMinSamplesSplit :: !Int rtMinLeafSize :: !Int rtMinImpurityDecrease :: !Double

Instances

Instances details

Show RegTreeConfig Source #

Instance details

Defined in DataFrame.DecisionTree.Regression

Methods

showsPrec :: Int -> RegTreeConfig -> ShowS #

show :: RegTreeConfig -> String #

showList :: [RegTreeConfig] -> ShowS #

Eq RegTreeConfig Source #

Instance details

Defined in DataFrame.DecisionTree.Regression

Methods

(==) :: RegTreeConfig -> RegTreeConfig -> Bool #

(/=) :: RegTreeConfig -> RegTreeConfig -> Bool #

Fit RegTreeConfig (Expr Double) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

Associated Types

type ModelOf RegTreeConfig (Expr Double)
Instance details Defined in DataFrame.DecisionTree.Model type ModelOf RegTreeConfig (Expr Double) = DecisionTreeRegressor
type FrameReq RegTreeConfig (Expr Double)
Instance details Defined in DataFrame.DecisionTree.Model type FrameReq RegTreeConfig (Expr Double) = 'AnyFrame

Methods

fit :: RegTreeConfig -> Expr Double -> FrameFor (Expr Double) -> FitResult (FrameFor (Expr Double)) (ModelOf RegTreeConfig (Expr Double)) Source #

type FrameReq RegTreeConfig (Expr Double) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

type FrameReq RegTreeConfig (Expr Double) = 'AnyFrame

type ModelOf RegTreeConfig (Expr Double) Source #

Instance details

Defined in DataFrame.DecisionTree.Model

type ModelOf RegTreeConfig (Expr Double) = DecisionTreeRegressor

defaultRegTreeConfig :: RegTreeConfig Source #

Fitted tree structure

data Tree a #

A fitted tree: a leaf value, or an internal node testing a boolean expression with True routing left.

Constructors

Leaf !a
Branch !(Expr Bool) !(Tree a) !(Tree a)

Instances

Instances details

Show a => Show (Tree a)
Instance details Defined in DataFrame.DecisionTree.Types Methods showsPrec :: Int -> Tree a -> ShowS # show :: Tree a -> String # showList :: [Tree a] -> ShowS #

treeDepth :: Tree a -> Int #

Solver configuration (fills `TreeConfig.linearSolverConfig`)

data SolverConfig #

Hyper-parameters for the FISTA solver.

Constructors

SolverConfig

Fields

scL1Lambda :: !Double
Strength of the L1 penalty on weights (intercept is not regularized).
scL2Lambda :: !Double
Strength of the L2 penalty (λ₂/2)·|w|² (Elastic Net; Zou & Hastie 2005). Combined with scL1Lambda this is the elastic-net objective; 0 reduces the solver to pure L1.
scMaxIter :: !Int
Maximum number of FISTA iterations.
scTol :: !Double
Convergence tolerance on the weight delta (L-inf norm).
scSampleWeights :: !(Maybe (Vector Double))
Optional per-row sample weights, length n (Nothing is uniform). Weights should have mean 1 (i.e. Σ w_i = N) so the Lipschitz bound stays valid; see fitLinearCandidate for the class-balanced construction.

Instances

Instances details

Show SolverConfig
Instance details Defined in DataFrame.LinearSolver Methods showsPrec :: Int -> SolverConfig -> ShowS # show :: SolverConfig -> String # showList :: [SolverConfig] -> ShowS #
Eq SolverConfig
Instance details Defined in DataFrame.LinearSolver Methods (==) :: SolverConfig -> SolverConfig -> Bool # (/=) :: SolverConfig -> SolverConfig -> Bool #

defaultSolverConfig :: SolverConfig #

Estimators

Classifier configuration

Instances

Instances

Instances

Regressor configuration

Instances

Fitted tree structure

Instances

Solver configuration (fills TreeConfig.linearSolverConfig)

Instances

Solver configuration (fills `TreeConfig.linearSolverConfig`)