有没有人知道呀

Title

    [D] data types -- Quick reference for data types


Description

    This entry provides a quick reference for data types allowed by Stata.  See [U] 12 Data for details.


Remarks

                                                         Closest to
    Storage                                              0 without
    type                 Minimum              Maximum    being 0     bytes
    ----------------------------------------------------------------------
    byte                    -127                  100    +/-1          1
    int                  -32,767               32,740    +/-1          2
    long          -2,147,483,647        2,147,483,620    +/-1          4
    float   -1.70141173319*10^38  1.70141173319*10^38    +/-10^-38     4
    double  -8.9884656743*10^307  8.9884656743*10^307    +/-10^-323    8
    ----------------------------------------------------------------------
    Precision for float  is 3.795x10^-8.
    Precision for double is 1.414x10^-16.


    String
    storage       Maximum
    type          length         Bytes
    ----------------------------------
    str1             1             1
    str2             2             2
     ...             .             .
     ...             .             .
     ...             .             .
    str244         244           244
    ----------------------------------

    Each element of data is said to be either type string or numeric.  The word "real" is sometimes used in place of numeric.
    Associated with each data type is a storage type.

    Strings are stored as str#, for instance, str1, str2, str3, ..., str244.  The number after the str indicates the maximum
    length of the string.  An str5 could hold the word "male", but not the word "female" because "female" has six characters.

    Numbers are stored as byte, int, long, float, or double, with the default being float.  byte, int, and long are said to be
    of integer type in that they can hold only integers.

    Stata keeps data in memory, and you should record your data as parsimoniously as possible.  If you have a string variable
    that has maximum length 6, it would waste memory to store it as a str20.  Similarly, if you have an integer variable, it
    would be a waste to store it as a double.


Precision of numeric storage types

    floats have about 7 digits of accuracy; the magnitude of the number does not matter.  Thus, 1234567 can be stored perfectly
    as a float, as can 1234567e+20.  The number 123456789, however, would be rounded to 123456792.  In general, this rounding
    does not matter.

    If you are storing identification numbers, the rounding could matter.  If the identification numbers are integers and take
    9 digits or less, store them as longs; otherwise, store them as doubles.  doubles have 16 digits of accuracy.

    Stata stores numbers in binary, and this has a second effect on numbers less than 1.  1/10 has no perfect binary
    representation just as 1/11 has no perfect decimal representation.  In float, .1 is stored as .10000000149011612.  Note
    that there are 7 digits of accuracy, just as with numbers larger than 1.  Stata, however, performs all calculations in
    double precision.  If you were to store 0.1 in a float called x and then ask, say, "list if x==.1", there would be nothing
    in the list.  The .1 that you just typed was converted to double, with 16 digits of accuracy (.100000000000000014...), and
    that number is never equal to 0.1 stored with float accuracy.

    One solution is to type "list if x==float(.1)".  The float() function rounds its argument to float accuracy; see [D]
    functions.  The other alternative would be store your data as double, but this is probably a waste of memory.  Few people
    have data that is accurate to 1 part in 10 to the 7th.  Among the exceptions are banks, who keep records accurate to the
    penny on amounts of billions of dollars.  If you are dealing with such financial data, store your dollar amounts as
    doubles.  See float().

先设置为double 录入就没有什么问题了

谢谢！

哲