Minor changes: Plotting, add __version__

README.md

@@ -119,8 +119,8 @@ pip install git+https://github.com/phiyodr/multilabel-oversampling
 ```
 
 
-## :construction_worker:
+## :construction_worker: Future work
 
-* [] Implement weighted sampling (so that samples which are already often in the new df are less often sampled)
+* [ ] Implement weighted sampling (so that samples which are already often in the new df are less often sampled)
 
 :sunflower:

multilabel_oversampling/__init__.py

@@ -1 +1,3 @@
+__version__ = "0.1.2"
+
 from .multilabel_oversampling import *

multilabel_oversampling/multilabel_oversampling.py

@@ -6,7 +6,7 @@
 from tqdm import tqdm
 import matplotlib.pyplot as plt
 import random
-import os
+import collections
 import math
 
 
@@ -137,14 +137,13 @@ def plot_results(self):
  """Plot target distribution before and after upsampling.
  Also plot the counts of each index-id.
  """
- plt.subplot(2,2,1)
+ plt.subplot(1,3,1)
  self.plot_distr(self.df, "before")
- plt.subplot(2,2,2)
+ plt.subplot(1,3,2)
  self.plot_distr(self.df_new, "after")
- plt.subplot(2,2,(3,4)) # MatplotlibDeprecationWarning
+ plt.subplot(1,3,3) # MatplotlibDeprecationWarning
  self.plot_index_counts(self.df_new)
  plt.tight_layout()
- plt.show()
  return plt
 
  def plot_distr(self, df, when):
@@ -153,17 +152,31 @@ def plot_distr(self, df, when):
  plt.title(f"Label distribution \n{when} upsampling")
  return plt
 
- def plot_index_counts(self, df_new):
+ def plot_individual_index_counts(self, df_new):
  """Plot upsampling counts for each index. 
 
  TODO make better xticks alignment"""
+ if df_new == None:
+ df_new == self.df_new
  idxs = list(df_new.index)
  lens = len(set(idxs))
  plt.hist(idxs, bins=lens, width=.1)#, edgecolor='k')
  xint = range(min(idxs), math.ceil(max(idxs))+1)
  plt.xticks(xint)
  plt.title("Draws per index\n in new df")
  return plt
+
+
+ def plot_index_counts(self, df_new=None):
+ if df_new is None:
+ df_new = self.df_new
+ x = list(collections.Counter(list(df_new.index)).values())
+ plt.hist(x, bins=max(x)+1, rwidth=.9)
+ plt.title("Frequency of indexes in df")
+ plt.xlabel('Frequency in dataset')
+ plt.ylabel('Counts')
+ return plt
+
 
 if __name__ == '__main__':
  seed_everything(seed=42)
@@ -172,4 +185,4 @@ def plot_index_counts(self, df_new):
  mlo = MultilabelOversampler(number_of_adds=100, plot=True)
  df_new = mlo.fit(df)
  print(mlo.df_new)
- mlo.plot_results()
+ mlo.plot_results()

setup.py

@@ -1,6 +1,6 @@
 import setuptools
-
-print(setuptools.find_packages())
+from multilabel_oversampling import __version__
+print(__version__)
 
 with open("README.md", "r") as file:
  long_description = file.read()
@@ -10,7 +10,7 @@
 
 setuptools.setup(
  name="multilabel-oversampling",
- version="0.1.1",
+ version=__version__, 
  author="Philipp J. Rösch",
  author_email="[email protected]",
  description="Multilabel Oversampling",