Merge pull request #222 from xuehuachunsheng/dev

修复rf-python3.6.py中文件打开没有关闭的问题

src/python/getting-started/digit-recognizer/rf-python3.6.py

@@ -19,7 +19,7 @@ import os.path
 import time
 
 # 数据路径
-data_dir = '/media/wsw/B634091A3408DF6D/data/kaggle/datasets/getting-started/digit-recognizer/'
+data_dir = '/Users/wuyanxue/Documents/GitHub/datasets/getting-started/digit-recognizer/'
 
 # 加载数据
 def opencsv():
@@ -57,8 +57,8 @@ def dRPCA(data, COMPONENT_NUM=100):
 def trainModel(X_train, y_train):
     print('Train RF...')
     clf = RandomForestClassifier(
-        n_estimators=140,
-        max_depth=20,
+        n_estimators=10,
+        max_depth=10,
         min_samples_split=2,
         min_samples_leaf=1,
         random_state=34)
@@ -99,16 +99,13 @@ def getModel(filename):
 # 结果输出保存
 def saveResult(result, csvName):
     i = 0
-    fw = open(csvName, 'w')
-    with open(os.path.join(data_dir, 'output/sample_submission.csv')
-              ) as pred_file:
+    n = len(result)
+    print('the size of test set is {}'.format(n))
+    with open(os.path.join(data_dir, 'output/Result_sklearn_RF.csv'), 'w') as fw:
         fw.write('{},{}\n'.format('ImageId', 'Label'))
-        for line in pred_file.readlines()[1:]:
-            splits = line.strip().split(',')
-            fw.write('{},{}\n'.format(splits[0], result[i]))
-            i += 1
-    fw.close()
-    print('Result saved successfully...')
+        for i in range(1, n + 1):
+            fw.write('{},{}\n'.format(i, result[i - 1]))
+    print('Result saved successfully... and the path = {}'.format(csvName))
 
 
 def trainRF():
@@ -151,7 +148,7 @@ def preRF():
     result = clf.predict(pcaPreData)
 
     # 结果的输出
-    saveResult(result,os.path.join(data_dir, 'output/Result_sklearn_rf.csv'))
+    saveResult(result, os.path.join(data_dir, 'output/Result_sklearn_rf.csv'))
     print("finish!")
     stopTime = time.time()
     print('PreModel load time used:%f s' % (stopTime - startTime))