Merge pull request #26 from behrica/generateTests

Auto generate tests

.gitignore

@@ -35,3 +35,6 @@ temp
 *.qmd
 book
 
+/.clj-kondo/
+test/noj_book/
+

build.clj

@@ -14,6 +14,8 @@
   (:require [clojure.tools.build.api :as b]
             [deps-deploy.deps-deploy :as dd]))
 
+
+
 (def lib 'org.scicloj/noj)
 (def version "2-alpha6")
 (def snapshot (str version "-SNAPSHOT"))
@@ -22,7 +24,7 @@
 (defn test "Run all the tests." [opts]
   (doseq [alias [:1.11 :1.12 :master]]
     (println "\nRunning tests for Clojure" (name alias))
-    (let [basis    (b/create-basis {:aliases [:test alias]})
+    (let [basis    (b/create-basis {:aliases (concat (opts :aliases [:test alias]))})
           cmds     (b/java-command
                     {:basis     basis
                      :main      'clojure.main
@@ -50,8 +52,19 @@
            :src-dirs  ["src"]
            :pom-data  (pom-template version))))
 
+(defn generate-tests [opts]
+  (let [basis    (b/create-basis {:aliases [:gen-tests :dev]})
+
+        cmds     (b/java-command
+                  {:basis     basis
+                   :main      'clojure.main
+                   :main-args ["-e" "(require '[gen-tests])(gen-tests/do-generate-tests)(System/exit 0)"]})
+        {:keys [exit]} (b/process cmds)]
+    (when-not (zero? exit) (throw (ex-info "Tests generation failed" {})))))
+(def opts {})
 (defn ci "Run the CI pipeline of tests (and build the JAR)." [opts]
-  (test opts)
+  (generate-tests (assoc opts :aliases [:dev]))
+  (test  (assoc opts :aliases [:dev :gen-tests :test]))
   (b/delete {:path "target"})
   (let [opts (jar-opts opts)]
     (println "\nWriting pom.xml...")
@@ -67,3 +80,7 @@
     (dd/deploy {:installer :remote :artifact (b/resolve-path jar-file)
                 :pom-file (b/pom-path (select-keys opts [:lib :class-dir]))}))
   opts)
+
+
+
+

build/gen_tests.clj

@@ -0,0 +1,18 @@
+(ns gen-tests
+  (:require
+   [scicloj.clay.v2.api :as clay]
+   [clojure.java.io :as io]
+   [clojure.string :as str]))
+
+(defn do-generate-tests []
+  (->>
+   (file-seq (io/file "notebooks/noj_book"))
+   (filter #(str/ends-with? (.toPath %) ".clj"))
+   (run!
+    #(let [p (.getAbsolutePath %)]
+       (println :generate-tests p)
+       (clay/make! {:source-path p
+                    :show false}))))
+  (shutdown-agents))
+
+(clay/make! {:source-path "notebooks/noj_book/automl.clj" :show false})

deps.edn

@@ -16,8 +16,13 @@
         org.scicloj/kind-pyplot {:mvn/version "1-beta1"}
         scicloj/clojisr {:mvn/version "1.0.0"}}
  :aliases
- {:build {:deps {io.github.clojure/tools.build {:mvn/version "0.9.6"}
-                 slipset/deps-deploy {:mvn/version "0.2.1"}}
+ {:gen-tests {:extra-paths ["build"]
+              :extra-deps { org.scicloj/clay {:mvn/version "2-beta16"}}}
+
+  :build {:deps {io.github.clojure/tools.build {:mvn/version "0.9.6"}
+                 slipset/deps-deploy           {:mvn/version "0.2.1"}}
+
+
           :ns-default build}
   :test {:extra-paths ["test" "notebooks"]
          :extra-deps {org.clojure/test.check {:mvn/version "1.1.1"}

notebooks/lr.clj

@@ -0,0 +1,44 @@
+(ns lr
+  (:require [tablecloth.api :as tc]
+            [tech.v3.tensor :as dt]
+            [scicloj.metamorph.ml :as ml]
+            [scicloj.ml.smile.regression]
+            [scicloj.metamorph.ml.loss :as loss]
+            [tech.v3.dataset.modelling :as dsmod]
+            [tech.v3.dataset.tensor :as dst]
+            [tech.v3.dataset :as ds]
+            [tech.v3.datatype.functional :as dtf]))
+
+(def x (dt/->tensor [[1 1] [1 2] [2 2] [2 3]]))
+(def y
+  (dtf/+ 3
+   (dt/reduce-axis x (fn [row] (dtf/dot-product row (dt/->tensor [1 2]))))))
+
+(def y-ds
+  (->
+   (dst/tensor->dataset x)
+   (tc/add-column :y y)
+   (dsmod/set-inference-target :y)))
+
+(def reg
+  (ml/train  y-ds
+             {:model-type :smile.regression/ordinary-least-square}))
+
+(- 1 (loss/rmse y (:y (ml/predict y-ds reg))));; => 1.0
+(seq (.coefficients (ml/thaw-model reg))) ;; => (2.9999999999999996 0.9999999999999998 2.0000000000000004)
+(seq (:y (ml/predict (tc/dataset [[3 5]]) reg)));; => (16.0)
+
+
+(defn Xy->dataset
+  "Combines tensor X and vector y to a dataset
+  where y is set to be the inference target
+  x need to be a tensor
+  y a seq or 1-D tensor
+  "
+  [x y]
+
+  (->
+   (dst/tensor->dataset x)
+   (ds/add-column (ds/new-column :y y))
+   (dsmod/set-inference-target :y)))
+

notebooks/noj_book/interactions_ols.clj

@@ -66,11 +66,11 @@ $$sales = b0 + b1 * youtube + b2 * facebook$$")
 (md "We have the following metrics:")
 (md "$RMSE$")
 (-> evaluations flatten first :test-transform :metric)
-(kindly/check = 0.933077510748531)
+(kindly/check = 1.772159024927988)
 
 (md "$R^2$")
 (-> evaluations flatten first :test-transform :other-metrices first :metric)
-(kindly/check = 0.9747551116991899)
+(kindly/check = 0.9094193687523886)
 
 (md "## Interaction effects")
 (md "Now we add interaction effects to it, resulting in this model equation:

notebooks/noj_book/ml_basic.clj

@@ -11,9 +11,13 @@
   (:require [tablecloth.api :as tc]
             [scicloj.metamorph.ml.toydata :as data]
             [tech.v3.dataset :as ds]
-            [scicloj.metamorph.ml :as ml]
+            [camel-snake-kebab.core :as csk]
+            [scicloj.kindly.v4.kind :as kind]
             [scicloj.kindly.v4.api :as kindly]))
 
+
+
+
 ;; ## Inspect data
 ;;
 ;;  The titanic data is part of `metamorph.ml` and in the form of a
@@ -91,8 +95,27 @@
    (ds-cat/fit-categorical-map relevant-titanic-data :pclass [0 1 2] :float64)
    (ds-cat/fit-categorical-map relevant-titanic-data :embarked ["S" "Q" "C"] :float64)])
 
+
 cat-maps
 
+(kind/test-last (fn [cat-maps]
+                  (every?
+                   true?
+                   (map
+                    #(.equals %1 %2)
+                    cat-maps
+                    [
+                     {:lookup-table {"male" 0, "female" 1},
+                      :src-column :sex,
+                      :result-datatype :float64}
+                     {:lookup-table {0 0, 1 1, 2 2, 3 3},
+                      :src-column :pclass,
+                      :result-datatype :float64}
+                     {:lookup-table {"S" 0, "Q" 1, "C" 2},
+                      :src-column :embarked,
+                      :result-datatype :float64}]))))
+
+
 ;; After the mappings are applied, we have a numeric dataset, as expected
 ;; by most models.
 (def numeric-titanic-data
@@ -103,6 +126,16 @@ cat-maps
 (tc/head
  numeric-titanic-data)
 
+
+(kind/test-last (fn [ds]
+                  (=
+                   [[0.0 3.0 0.0 0.0]
+                    [1.0 1.0 2.0 1.0]
+                    [1.0 3.0 0.0 1.0]
+                    [1.0 1.0 0.0 1.0]
+                    [0.0 3.0 0.0 0.0]]
+                   (ds/rowvecs ds))))
+
 ;; Split data into train and test set
 ;;  Now we split the data into train and test. By we use
 ;;  a :holdout strategy, so will get a single split in training an test data.
@@ -143,7 +176,7 @@ split
 (loss/classification-accuracy
  (:survived (ds-cat/reverse-map-categorical-xforms (:test split)))
  (:survived (ds-cat/reverse-map-categorical-xforms dummy-prediction)))
-;;  It's performance is poor, even worse the coin flip.
+;;  It's performance is poor, even worse then coin flip.
 
 
 ;; ## Logistic regression
@@ -166,7 +199,8 @@ split
  (:survived (ds-cat/reverse-map-categorical-xforms (:test split)))
  (:survived (ds-cat/reverse-map-categorical-xforms lreg-prediction)))
 
-;; Its performance is  better, 60 %
+(kind/test-last [= 0.7373737373737373])
+;; Its performance is  better, 73 %
 
 ;; ## Random forest
 ;; Next is random forest