Deleted files

[DilumAluthge]

I'm reorganizing some files. There are currently three files in the examples directory that we don't currently use. I'm deleting those files in #46.

So I'll put the contents of those files here.

We probably want to add these examples back to the documentation with the Literate + Documenter combination, similar to how I did the Bayesian linear regression in #46.

examples/example.jl

```julia using MLJ using MLJModelInterface using Soss using SossMLJ const MMI = MLJModelInterface m = @model X begin β ~ Cauchy() |> iid(2) yhat = X * β y ~ For(length(yhat)) do j Normal(yhat[j], 1) end end sm = SossModel(model=m) X = rand(Cauchy(),(10,2)) truth = rand(m(X=X)); truth.β truth.y mach = machine(sm, X, truth.y) fit!(mach) pred = MLJ.predict(mach, MMI.matrix(X)); rand.(pred) ```

examples/example.jmd

Loading some things ```julia using MLJ using MLJModelInterface using Soss using SossMLJ const MMI = MLJModelInterface ``` Now we make a Soss model ```julia; results="hidden" m = @model X begin β ~ Cauchy() |> iid(2) yhat = X * β y ~ For(length(yhat)) do j Normal(yhat[j], 1) end end ``` next we wrap this in a `SossMLJModel`, which connects us to MLJ. ```julia sm = SossMLJModel(model=m) ``` Let's get some fake data to try it out. ```julia; results="hidden" X = rand(Cauchy(),(10,2)) truth = rand(m(X=X)) ``` This gives us the parameter value ```julia truth.β ``` and response vector ```julia truth.y ``` Now we create and fit a machine: ```julia; results="hidden" mach = machine(sm, X, truth.y) fit!(mach) ``` And now for prediciton! Note that this is a probabilistic model, so `MLJ.predict` gives us a vector of sampleable results: ```julia pred = MLJ.predict(mach, X); rand.(pred) ``` And here's another sample: ```julia rand.(pred) ```

examples/example.md

Loading some things ````julia using MLJ using MLJModelInterface using Soss using SossMLJ const MMI = MLJModelInterface ```` Now we make a Soss model ````julia m = @model X begin β ~ Cauchy() |> iid(2) yhat = X * β y ~ For(length(yhat)) do j Normal(yhat[j], 1) end end ```` next we wrap this in a `SossModel` (note no space), which connects us to MLJ. ````julia sm = SossMLJModel(model=m) ```` ```` SossMLJModel( model = @model X begin β ~ Cauchy() |> iid(2) yhat = X * β y ~ For(length(yhat)) do j Normal(yhat[j], 1) end end , infer = Soss.dynamicHMC) @ 4…19 ```` Let's get some fake data to try it out. ````julia X = rand(Cauchy(),(10,2)) truth = rand(m(X=X)) ```` This gives us the parameter value ````julia truth.β ```` ```` 2-element Array{Float64,1}: 0.7795558910334254 1.3205090289240893 ```` and response vector ````julia truth.y ```` ```` 10-element Array{Float64,1}: -0.009656286618348886 0.9379274012136625 -1.0765641894263076 -4.884093727697979 0.44322739533204614 3.3290688788206877 9.804025505635714 -0.6967895342106162 -3.249175636559466 0.3924770024728915 ```` Now we create and fit a machine: ````julia mach = machine(sm, X, truth.y) fit!(mach) ```` And now for prediciton! Note that this is a probabilistic model, so `MLJ.predict` gives us a vector of sampleable results: ````julia pred = MLJ.predict(mach, X); rand.(pred) ```` ```` 10-element Array{Float64,1}: 0.6478518484624256 -0.09595465176171525 -2.4607605070347063 -6.263126184478209 1.0491468939935231 2.0356523077028923 9.115481356543857 -1.3853144490084879 -3.8689070474959615 0.6793707061491905 ```` And here's another sample: ````julia rand.(pred) ```` ```` 10-element Array{Float64,1}: -0.1191636821334261 0.5832262026481227 -3.5828613320656912 -6.423922375749305 -1.4119001531626445 2.3152820204955673 10.719893825466206 0.8946705351309647 -4.306828782379904 -1.4862297043387542 ````

[DilumAluthge]

I think those example files are superseded by the Bayesian linear regression example that we currently have in the docs.