Update docs

Author: juliohm

docs/src/related.md

@@ -25,14 +25,6 @@
   [IterableTables.jl](https://github.com/queryverse/IterableTables.jl).
   Similar to other alternatives above, the package is not intended for
   advanced statistical transforms.
-- [MLJ.jl](https://alan-turing-institute.github.io/MLJ.jl/dev/) is one
-  of the most popular packages for machine learning in Julia. The
-  package provides a facility for readily creating [non-branching
-  pipelines](https://alan-turing-institute.github.io/MLJ.jl/dev/linear_pipelines/#Linear-Pipelines)
-  which can include supervised learners, as well as the flexibility to
-  create more complicated composite machine learning models using
-  so-called [learning
-  networks](https://alan-turing-institute.github.io/MLJ.jl/dev/composing_models/#Learning-Networks). These composites
-  have the advantage that the hyper-parameters of the component models
-  appear as nested fields of the composite, which is useful in
-  hyper-parameter optimization.
+- [MLJ.jl](https://alan-turing-institute.github.io/MLJ.jl/dev/) is a
+  popular machine learning framework in Julia that adopts a different
+  design for pipelines based on mutability of structs.