Cleaning_ratings_votes_reviews.py

#A function to find all the non numeric values
def non_numerals(series):
  non_numerals = []
  for i in series.unique():
    try :
      i = float(i)
    except:
      non_numerals.append(i)
  return non_numerals

Cleaning_ratings_votes_reviews_2.py

# A function to replace the non-numeric values
def replace_nn_with(series, type_, fill_with = None, method = 'mean'):

  nn = non_numerals(series)
  print('-'*30)
  print('-'*30)
  print("Non Numerals in column ",series.name," : ",nn)

  series = series.replace(nn, np.nan, inplace = False)
  nulls = series.isnull().sum()
  if fill_with:
    series.fillna(fill_with, inplace = True)
    print("Filling Non Numerals with {}".format(fill_with))
  
  else:
    series = series.replace(nn, np.nan, inplace = False)

    if method == 'mean' :
      rep = series.astype(float).mean()
      print("Filling Non Numerals with MEAN = ", rep)

    elif method == 'median' :
      rep = series.astype(float).median()
      print("Filling Non Numerals with MEDIAN = ", rep)

    elif method == 'min' :
      rep = series.astype(float).min()
      print("Filling Non Numerals with MINIMUM = ", rep)

    else:
      print('Please pass a valid method as a string -- ("mean" or "median" or "min")')
      return 0

    series.fillna(rep, inplace = True)
    
  try:
    series = series.astype(type_)
    print(nulls, ": observations replaced")
    return series
  except:
    # Since type conversion of a string containting decimals to int is not possible, it is first converted to float
    series =  series.astype(float)
    print(nulls, ": observations replaced")
    series =  series.astype(type_)
    return series