TypeError: list indices must be integers, not float

I had this problem when using ANN and PyBrain on function testOnData().

So, I solved this problem putting "//" instead "/" inside the index on backprop.py source code.

I changed:

print(('Max error:', 
    max(ponderatedErrors), 
    'Median error:',
     sorted(ponderatedErrors)[len(errors) / 2])) # <-- Error area 

To:

print(('Max error:', 
    max(ponderatedErrors), 
    'Median error:',
     sorted(ponderatedErrors)[len(errors) // 2])) # <-- SOLVED. Truncated

I hope it will help you.

It looks like you are using Python 3.x. One of the important differences in Python 3.x is the way division is handled. When you do x / y, an integer is returned in Python 2.x because the decimal is truncated (floor division). However in 3.x, the / operator performs 'true' division, resulting in a float instead of an integer (e.g. 1 / 2 = 0.5). What this means is that your are now trying to use a float to reference a position in a list (e.g. my_list[0.5] or even my_list[1.0]), which will not work as Python is expecting an integer. Therefore you may first want to try using middle = (first + last) // 2, adjusting so that the result returns what you expect. The // indicates floor division in Python 3.x.

Kinda late to the party but you could also use:

middle = int((first + last) / 2)

In any case why you are getting your error is perfectly explained in RocketDonkey answer.

For your code to work you should also set:

position = binary_search(names, entered)

as Hugo Ferreira mentioned.

Also check this question: What is the difference between '/' and '//' when used for division?