Identification of spells or runs of similar values --------------------------------------------------
^spell^ [varname] [^if^ exp] [^in^ range] [ ^,^ { ^f^cond^(^fcondstr^)^ | ^c^ond^(^condstr^)^ | ^p^cond^(^pcondstr^)^ > } ^by(^byvarlist^) ce^nsor^(^censorvarlist^) e^nd^(^endvar^)^ ^s^eq^(^seqvar^) sp^ell^(^spellvar^) replace resort^ ]
Description -----------
^spell^ examines the data, treated as sequences or series, to identify spells or runs, and generates new variables:
(1) indicating distinct spells (0 for not in spell, or integers 1 up);
(2) giving sequence in spell (0 for not in spell, or integers 1 up); and
(3) indicating whether observations occur at the end of spells (0 or 1).
By default, these variables will be called ^_spell^, ^_seq^ and ^_end^.
Remarks -------
There are four ways of defining spells in ^spell^.
.-
First, given
^spell^ varname
a new spell starts whenever varname changes.
Second, a new spell starts whenever some condition defining the first observation in a spell is true. A spell ends just before a new spell starts. Such a condition may be specified by the ^fcond( )^ option. observation time 1 1 2 2 3 3 4 7 5 8 6 9
Suppose we wish to divide ^time^ into spells of consecutive values, in this case observations 1-3 and 4-6. A new spell starts when
^(time - time[_n - 1]) > 1^
which works for ^_n == 1^ as well because ^time[0]^ is treated as missing. Spells started by earthquakes, eruptions, accidents or other traumatic events may often be defined in this way.
Third, spells are defined by some condition being true for every observation in the spell. A spell ends when that condition becomes false. Such a condition may be specified by the ^cond( )^ option.
observation varname 1 0 2 10 3 11 4 12 5 0 6 13 7 0 8 14
Observations 2, 3, 4; 6; and 8 form 3 spells for the condition "varname > 0".
Observations 4; 6; and 8 form 3 spells for the condition "varname >= 12".
Fourth, a special but useful case of the previous kind is ^cond(^varname > 0 & varname < .^)^, that is, values of varname are positive (but not missing). Given daily data, spells of rain are defined by there being some rainfall every day. As a convenience, such conditions may be specified by ^pcond(^varname^)^, or more generally, ^pcond(^expression^)^.
.-
The order of the data when ^spell^ is called is taken to define position in the sequence. No explicit time or other ordering variable is required, but if any such exist, the user should ensure that data are in correct order before ^spell^ is called.
Spells are deemed to end at the last observation (under ^by( )^, at the last of each distinct group of observations).
Under ^if^ or ^in^, for identifying spells, observations are sorted so that all specified observations are placed together. See also ^resort^ below.
Missing values may be ignored by using ^if^ to exclude them. They are not ignored by default, as a convenience to users wishing to explore patterns of missing values. Recall that numeric missing ^.^ is treated as larger than any positive number. Thus be careful to exclude missing values where appropriate.
Conditions in terms of ^_n^ and ^_N^ are interpreted as follows. First, if ^if^ and/or ^in^ is specified, only those observations satisfying those conditions are considered, so that for example ^_N^ refers to the last such observation. Second, if ^by( )^ is specified, ^_n^ and ^_N^ refer to distinct groups defined by byvarlist. Third, ^spell^ respects the sort order of observations when it is invoked. These rules may operate together, in this order.
Two situations require particular care, because ^spell^ works best when the appropriate order of observations is defined unambiguously by a single criterion. The natural order of observations may depend on two or more variables, as when there are separate variables for year and month. The desired output may be sometimes best achieved by working with a combined variable which contains, say, year + (month - 0.5) / 12 or ^ym(^year^,^month^)^.
In some problems with tied values, ^spell^ may or may not place them in the same spell. The sequence variable will differ and the end variable may differ. There may be insufficient information to determine a unique sort order.
Given the earlier example, modified here,
observation time 1 1 2 2 3 2 <--- value changed from 3 4 7 5 8 6 9
and the same ^fcond( )^, which observations are put in positions 2 and 3 is arbitrary. By default, they will be assigned ^_seq^ of 2 and 3 and ^_end^ of 0 and 1, even though they are identical.
Options -------
^fcond(^fcondstr^)^ specifies a true or false condition that defines the start of a spell: to be precise, the first observation in a spell. A new spell starts whenever this condition is true. If varname is specified, and neither ^fcond( )^ nor ^cond( )^ is specified, ^fcond( )^ defaults to "varname != varname[_n-1]".
^cond(^condstr^)^ specifies a true or false condition that defines the spell.
^pcond(^pcondstr^)^ is equivalent to
^cond(^(pcondstr) > 0 & (pcondstr) < .)^)^
That is, some expression pcondstr evaluates to a positive number (but not missing). Commonly, the expression is just the name of a numeric variable.
Only one of ^fcond( )^, ^cond( )^ and ^pcond( )^ may be specified.
^by(^byvarlist^)^ specifies one or more variables that subdivide the data into groups. For example, the byvarlist could specify individual people. A further condition on a spell is then that the variable or variables in a byvarlist remain constant.
With ^by( )^, ^spell^ sorts data first by byvarlist and then by the order of the data when called. It then identifies spells. See also ^resort^ below.
^censor(^censorvarlist^)^ defines two new variables that are 1 if the spell is left- or right-censored and 0 otherwise. A left-censored spell starts at the first relevant observation (so it might have started earlier, except that we have no data to determine that). A right-censored spell ends at the last relevant observation (so it might have ended later, except that we have no data to determine that). ^censor( )^ should specify two new variable names, separated by white space (e.g. ^censor(cl cr)^), for indicating left-censored and right-censoring. As a special case, ^censor(.)^ indicates that _cl and _cr should be tried as new variable names.
^end(^endvar^)^ defines a new variable that is 1 at the end of each spell and 0 otherwise. ^_end^ is tried as a variable name if this option is not specified.
^seq(^seqvar^)^ defines a new variable that is the number of observations so far in the spell. ^_seq^ is tried as a variable name if this option is not specified.
^spell(^spellvar^)^ defines a new variable that is the number of spells so far. All observations in the first spell are tagged with 1, all in the second with 2, and so on. ^_spell^ is tried as a variable name if this option is not specified. Under ^by( )^, a separate count is kept for distinct values of byvarlist.
^replace^ indicates that any variables created by ^spell^ may overwrite existing variables with the same names.
^resort^ restores the original sort order of the observations after identifying spells. (The default if ^by( )^, ^if^ or ^in^ is invoked is to sort the data so that comparable and relevant observations are put together.)
Examples --------
. ^spell party^ Spells are distinct jobs (panel data):
. ^spell job, by(id)^
Number of spells (panel data):
. ^egen nspells = max(_spell), by(id)^
To define spells of consecutive values of ^time^: . ^spell, f(_n == 1 | (time - time[_n - 1]) > 1) by(id)^
Rainfall spells: . ^spell, p(rain)^
For spells in which rainfall was at least 10 mm every day: . ^spell, c(rain >= 10) e(hrend) s(hrseq)^
To get information on spell lengths (# observations):
. ^su hrseq if hrend^ . ^tab hrseq if hrend^
Length of each spell in a new variable:
. ^spell^ ... . ^egen length = max(_seq), by(_spell)^ . ^spell^ ... ^, by(id)^ . ^egen length = max(_seq), by(id _spell)^
Duration (length in time) of each spell in a new variable: . ^egen _tmax = max(time), by(_spell)^ . ^egen _tmin = min(time), by(_spell)^ . ^gen duration = _tmax - _tmin^
Cumulative totals of varname: . ^sort _spell _seq^ . ^qui by _spell : gen _tot = sum(^varname^) if _seq^
Sums of varname:
. ^egen total = sum(^varname^), by(_spell)^ Authors -------
Nicholas J. Cox, University of Durham, U.K. n.j.cox@@durham.ac.uk
Richard Goldstein richgold@@ix.netcom.com
Acknowledgements ----------------
Kit Baum, Stephen Jenkins and Fred Wolfe made very helpful comments. Philippe Van Kerm spotted a bug, which is fixed in 1.4.1. A question by Jan Dehn provoked some of the extensions in 1.5.0.