updating example.

Author: cnuernber

example/project.clj

@@ -4,7 +4,7 @@
   :license {:name "Eclipse Public License"
             :url "http://www.eclipse.org/legal/epl-v10.html"}
   :dependencies [[org.clojure/clojure "1.10.1-beta2"]
-                 [techascent/tech.ml "1.0-alpha3"]]
+                 [techascent/tech.ml "1.0-alpha4"]]
 
   :profiles {:dev {:dependencies [[org.clojure/tools.logging "0.3.1"]
                                   [ch.qos.logback/logback-classic "1.1.3"]]}})

example/src/tech/ml/example/classify.clj

@@ -4,9 +4,9 @@
             [tech.ml.dataset :as ml-dataset]
             [tech.ml.loss :as loss]
             [tech.libs.xgboost]
+            [tech.v2.datatype :as dtype]
             [tech.libs.smile.classification]
             [tech.libs.svm]
-            [clojure.core.matrix :as m]
             [taoensso.nippy :as nippy]
             [tech.io :as io]))
 
@@ -27,17 +27,23 @@
                                   (ml-dataset/ds-concat train-ds test-ds)
                                   train-ds)]
                    (->> base-systems
-                        (mapcat (fn [opts]
-                                  (let [gs-options (ml/auto-gridsearch-options opts)]
-                                    (println (format "Dataset: %s, Model: %s" name (:model-type opts)))
-                                    (->> (ml/gridsearch (merge options gs-options
-                                                               {:k-fold (if (> (first (m/shape train-ds)) 200)
-                                                                          5
-                                                                          3)
-                                                                :gridsearch-depth 75})
-                                                        loss/classification-loss
-                                                        train-ds)
-                                         (map (partial merge (dissoc base-dataset :test-ds :train-ds :options)))))))
+                        (mapcat
+                         (fn [opts]
+                           (let [gs-options (ml/auto-gridsearch-options opts)]
+                             (println (format "Dataset: %s, Model: %s"
+                                              name (:model-type opts)))
+                             (->> (ml/gridsearch
+                                   (merge options gs-options
+                                          {:k-fold (if (> (first (dtype/shape train-ds))
+                                                          200)
+                                                     5
+                                                     3)
+                                           :gridsearch-depth 75})
+                                   loss/classification-loss
+                                   train-ds)
+                                  (map (partial merge (dissoc base-dataset
+                                                              :test-ds :train-ds
+                                                              :options)))))))
                         (sort-by :average-loss)
                         (take 15)))})))))
 

example/src/tech/ml/example/svm_datasets.clj

@@ -5,6 +5,10 @@
             [tech.ml.dataset.svm :as svm]
             [tech.ml.dataset :as ds]
             [tech.ml.dataset.pipeline :as dsp]
+            [tech.ml.dataset.pipeline.pipeline-operators
+             :refer [without-recording
+                     pipeline-train-context
+                     pipeline-inference-context]]
             [tech.ml.dataset.column :as ds-col]
             [tech.v2.datatype :as dtype]))
 
@@ -36,11 +40,11 @@
   (time
    (let [{train-ds :dataset
           train-context :context}
-         (dsp/pipeline-train-context
+         (pipeline-train-context
           (-> (svm/parse-svm-file train-fname :label-map label-map)
               svm-pipeline))
          {test-ds :dataset} (when test-fname
-                              (dsp/pipeline-inference-context
+                              (pipeline-inference-context
                                train-context
                                (-> (svm/parse-svm-file test-fname :label-map label-map)
                                    svm-pipeline)))]