machine learning tutorial

book/basics/machine_learning.nim

@@ -0,0 +1,55 @@
+import nimib, nimibook
+
+nbInit(theme=useNimibook)
+
+nbText: md"""
+# 🤖👩‍🎓 Machine Learning
+
+In this tutorial we will cover the basics of [Machine Learning]
+in Nim by showing examples of the three classical tasks of
+[Clustering], [Classification], and [Regression]. 
+
+As reference dataset we will use [penguins dataset].
+An exploration of this dataset in Nim is available as an [example notebook] of nimib.
+Using this dataset we will:
+  - cluster penguins' features and see how much this matches species using [k-means clustering]
+  - classify penguins' sex and species starting from various sets of features using [logistic regression]
+  - predict the weight of penguins based on their bill size using [linear regression]
+
+Machine learning main concern is to build algorithms for automated data-driven prediction.
+The measure of success of such a modelling activity is usually encapsulated in some performance metric
+computed on specific subsets of data called [training, validation and test].
+In our examples we will:
+  - split the dataset in training and test
+  - process features appropriately
+  - fit the model on training set
+  - predict the test set using the trained model
+  - compute various metrics on predictions, appropriate for each of the above tasks
+  - validate our modelling approaches through [cross-validation]
+
+We will be using [arraymancer] for the implementation of machine learning algorithms,
+[datamancer] for manipulating data, [ggplotnim] for visualization.
+In Nim there is not (yet) a specific library that encapsulate
+machine learning concepts such as the `Estimator` of [scikit-learn].
+In this tutorial we will also try to build a simple api
+that could be a seed for a future [scinim/learn] library.
+
+[Machine Learning]: https://en.wikipedia.org/wiki/Machine_learning
+[Clustering]: https://en.wikipedia.org/wiki/Cluster_analysis
+[Classification]: https://en.wikipedia.org/wiki/Statistical_classification
+[Regression]: https://en.wikipedia.org/wiki/Regression_analysis
+[penguins dataset]: https://allisonhorst.github.io/palmerpenguins/
+[example notebook]: https://pietroppeter.github.io/nimib/penguins.html
+[logistic regression]: https://en.wikipedia.org/wiki/Logistic_regression
+[k-means clustering]: https://en.wikipedia.org/wiki/K-means_clustering
+[linear regression]: https://en.wikipedia.org/wiki/Linear_regression
+[training, validation and test]: https://en.wikipedia.org/wiki/Training,_validation,_and_test_sets
+[cross-validation]: https://en.wikipedia.org/wiki/Cross-validation_(statistics)
+[arraymancer]: https://github.com/mratsim/Arraymancer
+[datamancer]: https://github.com/scinim/datamancer
+[ggplotnim]: https://github.com/vindaar/ggplotnim
+[scikit-learn]: https://scikit-learn.org/stable/
+[scinim/learn]: https://github.com/scinim/learn
+"""
+
+nbSave

getting_started.nim

@@ -6,6 +6,7 @@ var book = newBookFromToc("SciNim Getting Started", "book"):
     entry("Common datatypes", "common_datatypes")
     entry("Data wrangling with dataframes", "data_wrangling")
     entry("Plotting", "basic_plotting")
+    entry("Machine Learning", "machine_learning")
     entry("Units", "units_basics")
   section("Numerical methods", "numerical_methods/index"):
     entry("Curve fitting", "curve_fitting")