Lathe.preprocess — Module

|====== Lathe.preprocess =====

|____________/ Generalized Processing ___________

|_____preprocess.TrainTestSplit

|_____preprocess.SortSplit

|_____preprocess.UniformSplit

|____________/ Feature Scaling ___________

|_____preprocess.Rescalar

|_____preprocess.ArbitraryRescale

|_____preprocess.MeanNormalization

|_____preprocess.StandardScalar

|____________/ Categorical Encoding ___________

|_____preprocess.OneHotEncoder

|_____preprocess.OrdinalEncoder

|_____preprocess.FloatEncoder

Lathe.preprocess.ArbitraryRescaler — Type

Arbitrary Rescaler

Description

Arbitrarily rescales an array.

Input

ArbitraryRescaler(x)

Positional Arguments

Array{Any} - x:: Array for which the original scaler should be based off of.

Output

scalar :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Applies the scaler to xt.

Data

a :: The minimum value in the array.

b :: The maximum value in the array.

Lathe.preprocess.FloatEncoder — Type

Float Encoder

Description

Float/Label Encodes an array.

Input

OneHotEncoder()

Output

encoder :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Returns an ordinally encoded xt.

Lathe.preprocess.MeanScaler — Type

Mean Normalizer

Description

Normalizes an array using the mean of the data.

Input

ArbitraryRescaler(x)

Positional Arguments

Array{Any} - x:: Array for which the original scaler should be based off of.

Output

scalar :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Applies the scaler to xt.

Data

a :: The minimum value in the array.

b :: The maximum value in the array.

avg :: The mean of the array.

Lathe.preprocess.OneHotEncoder — Type

OneHotEncoder

Description

One Hot Encodes a dataframe column into a dataframe.

Input

OneHotEncoder()

Output

encoder :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(df, symb) :: Applies the encoder to the dataframe key corresponding with symb on DF, then returns a dataframe with encoded results.

Lathe.preprocess.OrdinalEncoder — Type

Ordinal Encoder

Description

Ordinally Encodes an array.

Input

OrdinalEncoder(x)

Positional Arguments

Array{Any} - x:: Array for which the original scaler should be based off of.

Output

encoder :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Returns an ordinally encoded xt.

Lathe.preprocess.Rescaler — Type

Rescalar

Description

Rescales an array.

Input

Rescaler(x)

Positional Arguments

Array{Any} - x:: Array for which the original scaler should be based off of.

Output

scalar :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Applies the scaler to xt.

Data

min :: The minimum value in the array.

max :: The maximum value in the array.

Lathe.preprocess.StandardScaler — Type

Standard Scaler

Description

Normalizes an array using the z (Normal) distribution.

Input

StandardScaler(x)

Positional Arguments

Array{Any} - x:: Array for which the original scaler should be based off of.

Output

scalar :: A Lathe Preprocesser object.

Functions

Preprocesser.predict(xt) :: Applies the scaler to xt.

Data

dist :: Returns the normal distribution object for which this scaler uses.

Lathe.preprocess.SortSplit — Function

Sort Split

Description

Sorts an array, and then splits said array.

Input

SortSplit(x, .75, false)

Positional Arguments

Array{Any} - data:: The data to split.

Float64 - at:: A percentage that determines where the data is split.

Bool - rev:: Determines whether the order of the sort should be reversed.

Output

train:: The larger half of the split set.

test:: The smaller half of the split set.

Lathe.preprocess.TrainTestSplit — Function

TrainTestSplit

Description

Splits an array or dataframe into two smaller groups based on the percentage provided in the at parameter.

Input

TrainTestSplit(x, .75)

Positional Arguments

Array{Any}, DataFrame - data:: The data to split.

Float64 - at:: A percentage that determines where the data is split.

Output

train:: The larger half of the split set.

test:: The smaller half of the split set.

Lathe.preprocess.UniformSplit — Function

Uniform Split

Description

Uniform Split will split an array without shuffling the data first.

Input

UniformSplit(x, .75)

Positional Arguments

Array{Any} - data:: The data to split.

Float64 - at:: A percentage that determines where the data is split.

Output

train:: The larger half of the split set.

test:: The smaller half of the split set.

Lathe.preprocess.@onehot — Macro

OneHotEncodes a dataframe

Takes a symbol representing the column to one hot encode from and a DF.

df = (:A => ["hello","world"], :B => ["Foo", "Bar"])

encoded = @onehot df, :A