"How to aggregate age in polynominal type"

Hello, all

In my raw data set, I got age attribute as following example:

John, 64,

Alice, 33years,

Bob, 22years,

Mike, 50

So some of the value with a redundant 'years' at the end. What I eventually need is to check the average of age, and also group example into age group ( 0-9，10-19，20-29.and etc )

1) If i set the attribute type as integer when read file, then every example with redundant 'years' will only get as missing value. 

2) If I read the attribute as polynominal, and then use replace operator to remove the redudant part from attribute value. and then apply nominal to numeric, but still what I got is not a column with numeric type

is there a workaround for that?