get rid of some template stuff

more

Author: ablaom

Project.toml

@@ -1,12 +1,18 @@
-name = "MLJExampleInterface"
-uuid = "33837fe5-dbff-4c9e-8c2f-c5612fe2b8b6"
-authors = ["Anthony D. Blaom <[email protected]>"]
+name = "MLJText"
+uuid = "5e27fcf9-6bac-46ba-8580-b5712f3d6387"
+authors = ["Chris Alexander, Anthony D. Blaom <[email protected]>"]
 version = "0.1.0"
 
 [deps]
-Example = "7876af07-990d-54b4-ab0e-23690620f79a"
 MLJModelInterface = "e80e1ace-859a-464e-9ed9-23947d8ae3ea"
 ScientificTypesBase = "30f210dd-8aff-4c5f-94ba-8e64358c1161"
+TextAnalysis = "a2db99b7-8b79-58f8-94bf-bbc811eef33d"
+
+[compat]
+MLJModelInterface = "1.1.1"
+ScientificTypesBase = "1"
+TextAnalysis = "0.7.3"
+julia = "1"
 
 [extras]
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"

README.md

@@ -1,44 +1,8 @@
-# MLJExampleInterface.jl
+# MLJText.jl
 
-This repository is a template for creating repositories that contain
-glue code between packages providing machine learing algorithms, and
-the machine learning toolbox
-[MLJ](https://alan-turing-institute.github.io/MLJ.jl/dev/) - that is,
-for so-called *interface-only packages*.
-
-## When to use this template
-
-This template is intended for use when a package providing a machine
-learning model algorithm is not hosting the code that implements the
-MLJ model API, and a separate package for this purpose is to be
-created. This repo is itself a working implementation but should
-be used in conjunction with the more detailed [model implementation
-guidelines](https://alan-turing-institute.github.io/MLJ.jl/dev/adding_models_for_general_use/).
-
-## How to use this template
-
-1. Clone this repository or use it as a template if available from your organization. 
-
-2. Rename this repository, replacing the word "Example" with the name of the model-providing package.
-
-1. Develop the contents of src/MLJExampleInterface.jl appropriately
-
-2. Rename src/MLJExampleInterface.jl appropriately
-
-3. Remove Example from Project.toml and instead add the model-providing package
-
-3. **GENERATE A NEW UUID in Project.toml**
-
-4. Replace every instance of "Example" in this README.md with the name of the model-providing package and adjust the organization name in the link.
-
-5. Remove everything in this REAMDE.md except what is below the line you are currently reading 😉
-
-
-# MLJ.jl <--> Example.jl
-
-Repository implementing the [MLJ](https://alan-turing-institute.github.io/MLJ.jl/dev/) model interface for models provided by
-[Example.jl](https://github.com/JuliaLang/Example.jl).
+An [MLJ](https://alan-turing-institute.github.io/MLJ.jl/dev/)
+extension providing tools and model for text analysis.
 
 | Linux | Coverage |
 | :------------ | :------- |
-| [![Build Status](https://github.com/JuliaAI/MLJExampleInterface.jl/workflows/CI/badge.svg)](https://github.com/JuliaAI/MLJExampleInterface.jl/actions) | [![Coverage](https://codecov.io/gh/JuliaAI/MLJExampleInterface.jl/branch/master/graph/badge.svg)](https://codecov.io/github/JuliaAI/MLJExampleInterface.jl?branch=master) |
+| [![Build Status](https://github.com/JuliaAI/MLJText.jl/workflows/CI/badge.svg)](https://github.com/JuliaAI/MLJText.jl/actions) | [![Coverage](https://codecov.io/gh/JuliaAI/MLJText.jl/branch/master/graph/badge.svg)](https://codecov.io/github/JuliaAI/MLJText.jl?branch=master) |

src/MLJExampleInterface.jl renamed to src/MLJText.jl

@@ -1,11 +1,10 @@
-module MLJExampleInterface
+module MLJText
 
-# Example.jl doesn't actually provide machine learning models, so we
-# provide a module below with the same name to furnish us with simple
-# constant probabilistic classification. Note that `Example.fit`
-# ignores the training features `Xmatrix`.
+# The following is just boostrap code to get a working template. You
+# will remove this and replace "import .TextAnalysis" with "import
+# TextAnalysis" and any other deps you need.
 
-module Example
+module TextAnalysis
 
 function fit(Xmatrix::Matrix, yint::AbstractVector{<:Integer})
     classes = sort(unique(yint))
@@ -36,11 +35,11 @@ end # of module
 
 ### CONTINUATION OF TEMPLATE
 
-import .Example # substitute model-providing package name here (no dot)
+import .TextAnalysis # substitute model-providing package name here (no dot)
 import MLJModelInterface
 import ScientificTypesBase
 
-const PKG = "Example"          # substitute model-providing package name
+const PKG = "TextAnalysis"          # substitute model-providing package name
 const MMI = MLJModelInterface
 const STB = ScientificTypesBase
 
@@ -66,7 +65,7 @@ function MMI.fit(::CoolProbabilisticClassifier, verbosity, X, y)
     decode = MMI.decoder(y[1])                # for decoding int repr.
     classes_seen = decode(sort(unique(yint))) # ordered by int repr.
 
-    Θ = Example.fit(Xmatrix, yint)            # probability vector
+    Θ = TextAnalysis.fit(Xmatrix, yint)            # probability vector
     fitresult = (Θ, classes_seen)
     report = (n_classes_seen = length(classes_seen),)
     cache = nothing
@@ -79,7 +78,7 @@ function MMI.predict(::CoolProbabilisticClassifier, fitresult, Xnew)
     Xmatrix = MMI.matrix(Xnew)
 
     Θ, classes_seen = fitresult
-    prob_matrix = Example.predict(Xmatrix, Θ)
+    prob_matrix = TextAnalysis.predict(Xmatrix, Θ)
 
     # `classes_seen` is a categorical vector whose pool actually
     # includes *all* classes. The `UnivariateFinite` constructor
@@ -100,7 +99,7 @@ end
 MMI.metadata_pkg(CoolProbabilisticClassifier,
              name="$PKG",
              uuid="7876af07-990d-54b4-ab0e-23690620f79a",
-             url="https://github.com/JuliaLang/Example.jl",
+             url="https://github.com/JuliaLang/TextAnalysis.jl",
              is_pure_julia=true,
              license="MIT",
              is_wrapper=false

test/runtests.jl

@@ -1,4 +1,4 @@
-using MLJExampleInterface # substitute for correct interface pkg name
+using MLJTextAnalysisInterface # substitute for correct interface pkg name
 using Test
 using MLJBase
 import Distributions
@@ -10,7 +10,7 @@ rng = StableRNGs.StableRNG(123)
     p = 3
     nclasses = 5
     X, y = make_blobs(n, p, centers=nclasses, rng=rng);
-    model = MLJExampleInterface.CoolProbabilisticClassifier()
+    model = MLJTextAnalysisInterface.CoolProbabilisticClassifier()
     mach = machine(model, X, y)
     fit!(mach, rows=1:2, verbosity=0)
     yhat = predict(mach, rows=3)