Add _vp_sample function for sampling data with various data types

minjk-bl · minjk-bl · commit 37336349dd78 · 2022-05-20T16:49:10.000+09:00
diff --git a/js/m_visualize/Seaborn.js b/js/m_visualize/Seaborn.js
@@ -626,7 +626,8 @@ define([
                 code.appendFormatLine('plt.figure(figsize=({0}, {1}))', defaultWidth, defaultHeight);
                 if (useSampling) {
                     // data sampling code for preview
-                    convertedData = data + '.sample(n=' + sampleCount + ', random_state=0)';
+                    // convertedData = data + '.sample(n=' + sampleCount + ', random_state=0)';
+                    convertedData = com_util.formatString('_vp_sample({0}, {1})', data, sampleCount);
                     // replace pre-defined options
                     generatedCode = generatedCode.replaceAll(data, convertedData);
                 }   
diff --git a/python/variableCommand.py b/python/variableCommand.py
@@ -68,4 +68,19 @@ def _vp_get_profiling_list():
         result.append({ 'varName': v['varName'], 'title': title })
 
     return result
-    
+
+import numpy as _vp_np
+import random as _vp_rd
+def _vp_sample(data, sample_cnt):
+    dataType = type(data).__name__
+    sample_cnt = len(data) if len(data) < sample_cnt else sample_cnt
+
+    if dataType == 'DataFrame':
+        return data.sample(sample_cnt)
+    elif dataType == 'Series':
+        return data.sample(sample_cnt)
+    elif dataType == 'ndarray':
+        return data[_vp_np.random.choice(data.shape[0], sample_cnt, replace=False)]
+    elif dataType == 'list':
+        return _vp_rd.choices(data, k=sample_cnt)
+    return data
