The first thing I want to point out, you have an semantic error inside __init__, more specifically on self.experience. you had set self.experience = random.randint(self.min_experience, self.max.experience)

self.max.experience calls the fxn max() instead of the intended use of self.max_experience being a variable, be careful with certain naming techniques they can cause undo stress.

corrected:

import random

COLORS = ['yellow', 'red', 'blue', 'green']


class Monster(object):
  min_hit_points = 1
  max_hit_points = 1
  min_experience = 1
  max_experience = 1
  weapon = 'sword'
  sound = 'roar'

  def __init__(self, **kwargs):
    self.hit_points = random.randint(self.min_hit_points, self.max_hit_points)
    self.experience = random.randint(self.min_experience, self.max_experience)     
    self.color = random.choice(COLORS)                                                                    
    print(self.hit_points, self.experience, self.color)

    for key, value in kwargs.items():
      setattr(self, key, value)

  def battlecry(self):
    return self.sound.upper()

class Goblin(Monster):
  max_hit_points = 3
  max_experience = 2
  sound = 'squeak'

class Troll(Monster):
  min_hit_points = 3
  max_hit_points = 5
  min_experience = 2
  max_experience = 6
  sound = 'growl'

class Dragon(Monster):
  min_hit_points = 5
  max_hit_points = 10
  min_experience = 6
  max_experience = 10
  sound = 'raaaaaar'

if __name__ == '__main__':
  tim, gerald, drew, patty = Monster(), Goblin(), Troll(), Dragon()
  john = Goblin

Start by copying the above code and running it.

What is__init__?

Most if not all classes by default contain the __init__ function, it is commonly refereed to as a "class constructor". When a class is initialized the __init__ fxn whether you overrode the fxn or not is run. It constructs the class on initialization....for example:

The output of the above code:

1 1 blue
1 2 blue
4 4 green
7 7 green

You can run this code multiple times and the output could be different every time thanks to __init__. Did you notice that john = Goblin didn't have an output?(explanation below)

How does this pertain to the code above?

Above we set gerald = Goblin() which is different from when we set john = Goblin. Lets call gerald case#1 and john case#2, in case#1 the class is initialized and in case#2 just the opposite. If you tried to call john.color since the class wasn't initialized python would throw an error AttributeError: type object 'Goblin' has no attribute 'color' on the flip side gerald.color would return a randomly generated value from the list COLORS. However you can still initialize case#2 by john(), at this point __init__ is called.

Class inheritance

Monster is what's called a super class or parent class, values in it will be inherited by the classes Dragon, Troll, and Goblin.

Dragon, Troll, and Goblin are referred to as a child class or a child of Monster since they inherit certain attributes from Monster.

**kwargs

**kwargs is a place holder variable that allows you to pass an undetermined number of variables to it. For instance since you aren't passing minimum values to the parent class from Goblin, you use **kwargs to prevent TypeError $FUNCTION() missing $NUM required positional arguments from occurring. By using **kwargs you are saying you are going to pass an indiscriminate number of variables to the function.

What's the point of using init and **kwargs if we are manually assigning so many attributes to these classes/subclasses anyways?

If you notice, you aren't assigning direct values to your classes you are passing parameters to the parent class which are calculated by the fxn __init__ upon initialization.

I hope I have answered your question, please let me know if you need further clarification.

Recommended reading:

• Fundamentals of Python: First Programs by: Kenneth A. Lambert

https://www.amazon.com/Fundamentals-Python-Programs-Introduction-Programming/dp/1111822700/ref=sr_1_1?ie=UTF8&qid=1470645403&sr=8-1&keywords=fundamentals+of+python+first+programs

-Fundamentals of Python: Data Structures by: Kenneth A. Lambert

https://www.amazon.com/Fundamentals-Python-Structures-Kenneth-Lambert/dp/1285752007/ref=pd_sim_14_1?ie=UTF8&dpID=51MD0E6TRSL&dpSrc=sims&preST=_AC_UL160_SR129%2C160_&psc=1&refRID=HYEFA3046R5ETBFQV87P

• Functional Programming by Bruce Maclennan

https://www.amazon.com/Functional-Programming-Practice-Bruce-Maclennan/dp/0201137445/ref=sr_1_10?s=books&ie=UTF8&qid=1470645482&sr=1-10&keywords=functional+programming

• Clean Code: A Handbook of Agile Software Craftsmanship by Robert C. “Uncle Bob” Martin

https://www.amazon.com/Clean-Code-Handbook-Software-Craftsmanship/dp/0132350882/ref=sr_1_1?s=books&ie=UTF8&qid=1470645557&sr=1-1&keywords=clean+code