consistently reorder dataset

lapolonio · lapolonio · commit c01d8df3fedf · 2019-10-21T01:26:13.000-04:00
diff --git a/step_1/bert/run_classifier.py b/step_1/bert/run_classifier.py
@@ -317,7 +317,7 @@ def load_dataset(self, directory, set_type):
     pos_df["polarity"] = "positive"
     neg_df["polarity"] = "negative"
     return self._create_examples(
-      pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),
+      pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),
       set_type
     )
 
diff --git a/step_2/bert/run_classifier.py b/step_2/bert/run_classifier.py
@@ -325,7 +325,7 @@ def load_dataset(self, directory, set_type):
     pos_df["polarity"] = "positive"
     neg_df["polarity"] = "negative"
     return self._create_examples(
-      pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),
+      pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),
       set_type
     )
 
diff --git a/step_3/bert/run_classifier.py b/step_3/bert/run_classifier.py
@@ -325,7 +325,7 @@ def load_dataset(self, directory, set_type):
     pos_df["polarity"] = "positive"
     neg_df["polarity"] = "negative"
     return self._create_examples(
-      pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),
+      pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),
       set_type
     )
 
diff --git a/step_4/bert/run_classifier.py b/step_4/bert/run_classifier.py
@@ -325,7 +325,7 @@ def load_dataset(self, directory, set_type):
     pos_df["polarity"] = "positive"
     neg_df["polarity"] = "negative"
     return self._create_examples(
-      pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),
+      pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),
       set_type
     )
 

Original file line number	Diff line number	Diff line change
`@@ -317,7 +317,7 @@ def load_dataset(self, directory, set_type):`
`317`	`317`	`pos_df["polarity"] = "positive"`
`318`	`318`	`neg_df["polarity"] = "negative"`
`319`	`319`	`return self._create_examples(`
`320`		`- pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),`
	`320`	`+ pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),`
`321`	`321`	`set_type`
`322`	`322`	`)`
`323`	`323`
Original file line number	Diff line number	Diff line change
`@@ -325,7 +325,7 @@ def load_dataset(self, directory, set_type):`
`325`	`325`	`pos_df["polarity"] = "positive"`
`326`	`326`	`neg_df["polarity"] = "negative"`
`327`	`327`	`return self._create_examples(`
`328`		`- pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True),`
	`328`	`+ pd.concat([pos_df, neg_df]).sample(frac=1, random_state=42).reset_index(drop=True),`
`329`	`329`	`set_type`
`330`	`330`	`)`
`331`	`331`