Python function not finding global variable

Question

I am creating a program to calculate b1 for a linear regression program in Python.

When trying to assign a value to my b1 global variable, my IDE doesn't find it. You can find the variable assignment at the last line of my function.

x_data = [4, 8, 6, 3, 4, 6, 2, 1, 3]
y_data = [2, 4, 7, 3, 4, 8, 3, 2, 5]
b1 = 0
def b0_b1_calc():
 x_sum = 0
 y_sum = 0
 for i in x_data:
 x_sum += x_data[i]
 y_sum += y_data[i]
 x_mean = round(1, x_sum / len(x_data))
 y_mean = round(1, y_sum / len(y_data))
 x_list = []
 y_list = []
 for i in x_data:
 x_list.append(i - x_mean)
 y_list.append(i - y_mean)
 x_minus_x_squared = []
 for i in x_list:
 x_minus_x_squared.append((x_list[i] - x_mean) ** 2)
 x_sum = sum(x_minus_x_squared)
 x_y = []
 for i in x_data:
 x_y.append(y_list * x_list)
 x_y = sum(x_y)
 b1 = x_y / x_sum
 
print(str(b1))
 

• matszwecja: 1. Try to use global variables as little as possible. 2. If you really do NEED to use global, mark the variable as such with global keyword
• mnikley: In your last line of your function, you create a new variable b1 which is only available inside the function. If you want to access that value later, you have to: 1) call your function, 2) either print the result inside the function or return the result. I would advise you against global variables here, simply change your last line in the function to return return x_y / x_sum, and secondly replace your statement print(str(b1)) with print(b0_b1_calc()))
• Jussi Nurminen: The normal way to do this would be to pass in data as function arguments, and return results from function by return. Why don't you do that?

Answer 1

You have to use global to access global variables. There were a few more errors in the code that I have fixed and annotated too:

def b0_b1_calc():
 # Use global to access global variables
 global b1
 # sum() summates a list
 x_sum = sum(x_data)
 y_sum = sum(y_data)
 # This was the wrong way around
 x_mean = round(x_sum / len(x_data))
 y_mean = round(y_sum / len(y_data))
 x_list = [] # could also be [i - x_mean for i in x_data]
 y_list = [] # could also be [i - y_mean for i in x_data]
 for i in x_data:
 x_list.append(i - x_mean)
 y_list.append(i - y_mean)
 # could also be [(x_list[i] - x_mean) ** 2 for i in x_list]
 x_minus_x_squared = []
 for i in x_list:
 x_minus_x_squared.append((x_list[i] - x_mean) ** 2)
 x_sum = sum(x_minus_x_squared)
 x_y = []
 for i in x_data:
 # could be simplified to [np.array(y_list) * np.array(x_list) for i in x_data]
 # use np.array for vectorial array multiplication
 x_y.append(np.array(y_list) * np.array(x_list))
 
 x_y = sum(x_y)
 b1 = x_y / x_sum
b0_b1_calc()

The preferred way to get around globals is to call and return:

def b0_b1_calc():
 # ... code here
 return b1
b1 = b0_b1_calc()

• Freddy Mcloughlan: Hold on, the code has some errors in it. Fixing now
• Freddy Mcloughlan: Let me know if that answer doesn't work

Answer 2

Please declare b1 as global in the function as following:

 def b0_b1_calc():
 global b1
 # remaining code in the function