minor cleaning

app/appraiser.py

@@ -1,4 +1,3 @@
-from numpy.typing import ArrayLike
 from typing import List
 
 class Appraiser():

app/baselines/datascope_wrapper.py

@@ -3,7 +3,7 @@
 from datascope.importance.common import SklearnModelAccuracy
 from datascope.importance.shapley import ShapleyImportance, ImportanceMethod
 from appraiser import Appraiser
-from classifier import LogisticClassifier as Classifier
+from classifier import XGBClassifier as Classifier
 
 utility_pipeline = pipeline.make_pipeline(
     Classifier()

app/baselines/debugging.py

@@ -1,6 +1,6 @@
 import numpy as np
 from sklearn.model_selection import train_test_split
-from classifier import LogisticClassifier as Classifier
+from classifier import XGBClassifier as Classifier
 from appraiser import Appraiser
 
 class CustomAppraiser(Appraiser):

app/baselines/influence_function.py

@@ -1,5 +1,6 @@
 import torch
 from appraiser import Appraiser
+# we use LogisticClassifier here as a proxy
 from classifier import LogisticClassifier as Classifier
 from baselines.ptif.calc_influence_function import calc_img_wise
 from torch.utils.data import TensorDataset, DataLoader

app/baselines/random_pick.py

@@ -1,5 +1,4 @@
 import numpy as np
-from sklearn.model_selection import train_test_split
 from appraiser import Appraiser
 
 class RandomAppraiser(Appraiser):

app/classifier.py

@@ -1,13 +1,10 @@
-from loguru import logger
 import numpy as np
 from pyarrow import parquet as pq
 import sklearn
 import torch
 import xgboost
 import torch.nn as nn
 from torch.utils.data import TensorDataset, DataLoader
-from sklearn.linear_model import LogisticRegressionCV
-
 
 class XGBClassifier():
     def __init__(self):

app/plotter_overall.py

@@ -29,8 +29,6 @@ def aggregate_data(evaluation_file, result_folder):
     return data
 
 def plot(data, result_folder, score_metric='auc'):
-    data = data[data['method']!='my_debug']
-
     sns.set(
         font="DejaVu Sans",
         context="paper",

app/utils.py

@@ -1,7 +1,7 @@
 import numpy as np
 import pyarrow as pa
-from sklearn import metrics
 from loguru import logger
+
 def fix(proposed_fixes, train, budget, gt_df):
     if len(proposed_fixes) > budget:
         raise ValueError("Submission takes more budget than expected, {}>{}".format(len(proposed_fixes), budget))