Added __str__ methods, and additional check when CorrelatedFeatureRem…

…over loads data from disk

src/removers.py

@@ -1,5 +1,6 @@
 import pandas as pd
 import numpy as np
+from .exceptions import DataProcessorError
 
 
 class BaseFeatureRemover:
@@ -27,13 +28,21 @@ def __init__(self, correlation_threshold=0.9, verbose=True, force_recompute=Fals
         self.fitted = False
         self.verbose = verbose
         self.force_recompute = force_recompute
+        self.persistent = True
         self.write_to_file = write_to_file
         self.load_from_file = load_from_file
 
+    def __str__(self):
+        return 'CorrelatedFeatureRemover(correlation_threshold={})'.format(self.correlation_threshold)
+
     def fit(self, df, feature_columns):
         if self.load_from_file:
             corr = pd.read_csv(self.load_from_file, index_col=0)
             corr = corr.abs()
+
+            for col in feature_columns:
+                if col not in corr.columns:
+                    raise DataProcessorError("Column '{}' not found in correlation file".format(col))
         else:
             corr = df[feature_columns].corr()
             if self.write_to_file:
@@ -53,24 +62,23 @@ def fit(self, df, feature_columns):
 
 
 class AlmostConstantFeatureRemover:
-    def __init__(self, max_count_percent=90, verbose=True, force_recompute=False, write_to_file=False,
-                 load_from_file=False):
+    def __init__(self, max_count_percent=90, verbose=True, force_recompute=False):
         """
         If a column has a single value that makes up more than max_count_percent of the values, remove it
         :param max_count_percent:
         :param verbose:
         :param force_recompute:
-        :param write_to_file:
-        :param load_from_file:
         """
         self.max_count_percent = max_count_percent
         self.columns_to_remove = []
         self.columns_to_leave = []
         self.fitted = False
         self.verbose = verbose
         self.force_recompute = force_recompute
-        self.write_to_file = write_to_file
-        self.load_from_file = load_from_file
+        self.persistent = False
+
+    def __str__(self):
+        return 'AlmostConstantFeatureRemover(max_count_percent={})'.format(self.max_count_percent)
 
     def fit(self, df, feature_columns):
         len_df = len(df)