The default float converter, defined in pymysql.converters.escape_float looks like this:

def escape_float(value, mapping=None):
    return ('%.15g' % value)

For floating point values with more than 15 digits, this loses information. The mysqlclient equivalent preserves it:

def Float2Str(o, d):
    s = repr(o)
    if s in ('inf', 'nan'):
        raise ProgrammingError("%s can not be used with MySQL" % s)
    if 'e' not in s:
        s += 'e0'
    return s

Losing this precision can arise, for example, in sending the output of time.time() to the database, and sometimes those final digits make a difference. Here's one example, using a timestamp of 1564063129.1277142:

$ python3.7 -c 'from MySQLdb.converters import Float2Str; print(Float2Str(1564063129.1277142, None))'
1564063129.1277142e0
$ python3.7 -c 'from pymysql.converters import escape_float; print(escape_float(1564063129.1277142))'
1564063129.12771

Notice PyMySQL lops off the final two digits.

This was easy enough to fix by overriding the conv dict to the connect() method, but maybe the default could be to preserve all the digits, like MySQLdb does?