.to_f is the right way.
Example:
irb(main):001:0> "-10".to_f
=> -10.0
irb(main):002:0> "-10.33".to_f
=> -10.33
Maybe your string does not include a regular “-” (dash)? Or is there a space between the dash and the first numeral?
Added:
If you know that your input string is a string version of a floating number, eg, “10.2”, then .to_f is the best/simplest way to do the conversion.
If you’re not sure of the string’s content, then using .to_f will give 0 in the case where you don’t have any numbers in the string. It will give various other values depending on your input string too. Eg
irb(main):001:0> "".to_f
=> 0.0
irb(main):002:0> "hi!".to_f
=> 0.0
irb(main):003:0> "4 you!".to_f
=> 4.0
The above .to_f behavior may be just what you want, it depends on your problem case.
Depending on what you want to do in various error cases, you can use Kernel::Float as Mark Rushakoff suggests, since it raises an error when it is not perfectly happy with converting the input string.