debug sparse

scicloj · Oct 1, 2024 · 9a0fd7a · 9a0fd7a
1 parent 49c906f
commit 9a0fd7a
Show file tree

Hide file tree

Showing 2 changed files with 39 additions and 16 deletions.
diff --git a/src/scicloj/ml/xgboost.clj b/src/scicloj/ml/xgboost.clj
@@ -174,16 +174,17 @@ subsample may be set to as low as 0.1 without loss of model accuracy. Note that
 
 (defn- sparse->labeled-point [^SparseArray sparse target n-sparse-columns]
   (let [x-i-s
-        (map
+        (mapv
          #(hash-map :i  (.i ^SparseArray$Entry %) :x (.x ^SparseArray$Entry %))
          (iterator-seq
           (.iterator sparse)))]
+    (println :x-i-s (count x-i-s))
     (LabeledPoint. target
                    n-sparse-columns
                    (into-array Integer/TYPE (map :i x-i-s))
                    (into-array Float/TYPE (map :x x-i-s)))))
 
-(defn- sparse-feature->dmatrix [feature-ds target-ds sparse-column n-sparse-columns]
+(defn sparse-feature->dmatrix [feature-ds target-ds sparse-column n-sparse-columns]
   (DMatrix.
    (.iterator
     ^Iterable (map
@@ -194,6 +195,7 @@ subsample may be set to as low as 0.1 without loss of model accuracy. Note that
    nil))
 
 
+
 (defn- dataset->labeled-point-iterator
   "Create an iterator to labeled points from a possibly quite large
   sequence of maps.  Sets expected length to length of first entry"
@@ -297,11 +299,14 @@ c/xgboost4j/java/XGBoost.java#L208"))
           params (->>  cleaned-options
                       ;;Adding in some defaults
                        (merge
-                        {:alpha 0.0
-                         :eta 0.3
-                         :lambda 1.0
-                         :max-depth 6
-                         :subsample 0.87}
+                        {
+                        ;;  :alpha 0.0
+                        ;;  :eta 0.3
+                        ;;  :lambda 1.0
+                        ;;  :max-depth 6
+                        ;;  :subsample 0.87
+
+                         }
 
                         cleaned-options
                         (when label-map

diff --git a/test/scicloj/ml/xgboost_test.clj b/test/scicloj/ml/xgboost_test.clj
@@ -205,16 +205,33 @@
                                (dtype/emap val-map :keyword col))))
          (ds/categorical->number cf/categorical)
          (ds-mod/set-inference-target :Score)))
-
+
+
+
+      (def feature-ds
+        (cf/feature reviews))
+
+      (def target-ds
+        (cf/target reviews))
+
+      (require '[scicloj.ml.xgboost])
+      (def d
+        (scicloj.ml.xgboost/sparse-feature->dmatrix 
+         feature-ds
+         target-ds
+         :bow-sparse
+         1000
+         ))
       (def trained-model
         (ml/train reviews {:model-type :xgboost/classification
                            :sparse-column :bow-sparse
-                           :n-sparse-columns 100
-                           :silent 0
-                           :round 1
-                           :eval-metric "merror"
-                           :watches {:test-ds (ds/sample  reviews 10)}}))
-
+                           :n-sparse-columns 1000
+                           ;:silent 0
+                           ;:round 1
+                           ;:eval-metric "merror"
+                           ;:watches {:test-ds (ds/sample  reviews 10)}
+                           }))
+
 
 
       (def prediction
@@ -227,5 +244,6 @@
         (ml/train-k-fold reviews {:model-type :xgboost/classification
                                   :sparse-column :bow-sparse}))
 
-
-      (ml/explain folds))
+
+      (ml/explain folds)
+      )