Notebook: https://github.com/dsaharia/Data-Science-Study-Group/blob/master/ml-introduction/grad-des-exdata.ipynb

Notes and suggestions for experimentation and further reading:

Note 1

x_population = data[0].values  # Extracting the first column and storing in a pandas series.
y_profits = data[1].values # Same for the second column

Don't need to store pandas columns into separate variables. You can rename the columns to population and profits. See this. And then you can refer to columns as data['population']
etc.

Note 2

The use of range in multiple places, example:

    for i in range(0,len(x_population)):
        x = x_population[i]
        y = y_profits[i]
        total_error += ((m * x + b) - y)**2

Range is useful too in certain cases, but also try enumerate. Something like

for popvar, popval in enumerate(x_population):
      print(popvar, popval)

See what you get. This way, you don't need the len(x_population).

Note 3 Initialization

total_error = 0 doesn't needs to be initialized. If you want to store all total_error values, you can initialize using total_error=[] and then append the value in every loop. Check how to append values in a loop.

P.S. Plot the cost (y-axis) with iterations (x-axis). You will have a better handle over the evolution.