[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Pspp-cvs] pspp src/language/stats/ChangeLog src/language/...
|
From: |
Ben Pfaff |
|
Subject: |
[Pspp-cvs] pspp src/language/stats/ChangeLog src/language/... |
|
Date: |
Thu, 02 Aug 2007 03:07:18 +0000 |
CVSROOT: /cvsroot/pspp
Module name: pspp
Changes by: Ben Pfaff <blp> 07/08/02 03:07:18
Modified files:
src/language/stats: ChangeLog frequencies.q
tests : ChangeLog
tests/command : weight.sh
tests/stats : percentiles-compatible.sh
percentiles-enhanced.sh
Log message:
Clean up handling of median, by treating it almost like any other
percentile. Fixes bug #17424. Thanks to John Darrington for
review.
* frequencies.q (internal_cmd_frequencies): Fix handling of bit
masks for `stats' variable. If median is selected, turn it off
and add a 50th percentile.
(add_percentile): Simplify code a little.
(calc_stats): Drop special casing of median.
(dump_statistics): Ditto, except that we label the 50th percentile
as "50 (Median)" to make it clear that it's also the median.
CVSWeb URLs:
http://cvs.savannah.gnu.org/viewcvs/pspp/src/language/stats/ChangeLog?cvsroot=pspp&r1=1.58&r2=1.59
http://cvs.savannah.gnu.org/viewcvs/pspp/src/language/stats/frequencies.q?cvsroot=pspp&r1=1.34&r2=1.35
http://cvs.savannah.gnu.org/viewcvs/pspp/tests/ChangeLog?cvsroot=pspp&r1=1.100&r2=1.101
http://cvs.savannah.gnu.org/viewcvs/pspp/tests/command/weight.sh?cvsroot=pspp&r1=1.19&r2=1.20
http://cvs.savannah.gnu.org/viewcvs/pspp/tests/stats/percentiles-compatible.sh?cvsroot=pspp&r1=1.10&r2=1.11
http://cvs.savannah.gnu.org/viewcvs/pspp/tests/stats/percentiles-enhanced.sh?cvsroot=pspp&r1=1.10&r2=1.11
Patches:
Index: src/language/stats/ChangeLog
===================================================================
RCS file: /cvsroot/pspp/pspp/src/language/stats/ChangeLog,v
retrieving revision 1.58
retrieving revision 1.59
diff -u -b -r1.58 -r1.59
--- src/language/stats/ChangeLog 1 Aug 2007 03:33:15 -0000 1.58
+++ src/language/stats/ChangeLog 2 Aug 2007 03:07:17 -0000 1.59
@@ -1,3 +1,16 @@
+2007-08-01 Ben Pfaff <address@hidden>
+
+ Clean up handling of median, by treating it almost like any other
+ percentile. Fixes bug #17424. Thanks to John Darrington for
+ review.
+ * frequencies.q (internal_cmd_frequencies): Fix handling of bit
+ masks for `stats' variable. If median is selected, turn it off
+ and add a 50th percentile.
+ (add_percentile): Simplify code a little.
+ (calc_stats): Drop special casing of median.
+ (dump_statistics): Ditto, except that we label the 50th percentile
+ as "50 (Median)" to make it clear that it's also the median.
+
2007-07-31 Ben Pfaff <address@hidden>
Remove integer mode from FREQUENCIES and incidentally fix bug
Index: src/language/stats/frequencies.q
===================================================================
RCS file: /cvsroot/pspp/pspp/src/language/stats/frequencies.q,v
retrieving revision 1.34
retrieving revision 1.35
diff -u -b -r1.34 -r1.35
--- src/language/stats/frequencies.q 1 Aug 2007 03:33:15 -0000 1.34
+++ src/language/stats/frequencies.q 2 Aug 2007 03:07:17 -0000 1.35
@@ -156,8 +156,6 @@
static struct percentile *percentiles;
static int n_percentiles;
-static int implicit_50th ;
-
/* Groups of statistics. */
#define BI BIT_INDEX
#define frq_default \
@@ -323,14 +321,14 @@
if (cmd.a_statistics[FRQ_ST_ALL])
stats |= frq_all;
if (cmd.sort != FRQ_AVALUE && cmd.sort != FRQ_DVALUE)
- stats &= ~frq_median;
+ stats &= ~BIT_INDEX (frq_median);
for (i = 0; i < frq_n_stats; i++)
if (cmd.a_statistics[st_name[i].st_indx])
stats |= BIT_INDEX (i);
if (stats & frq_kurt)
- stats |= frq_sekurt;
+ stats |= BIT_INDEX (frq_sekurt);
if (stats & frq_skew)
- stats |= frq_seskew;
+ stats |= BIT_INDEX (frq_seskew);
/* Calculate n_stats. */
n_stats = 0;
@@ -363,7 +361,14 @@
add_percentile (j / (double) cmd.n_ntiles[i]);
}
}
-
+ if (stats & BIT_INDEX (frq_median))
+ {
+ /* Treat the median as the 50% percentile.
+ We output it in the percentiles table as "50 (Median)." */
+ add_percentile (0.5);
+ stats &= ~BIT_INDEX (frq_median);
+ n_stats--;
+ }
/* Do it! */
input = casereader_create_filter_weight (proc_open (ds), dataset_dict (ds),
@@ -876,11 +881,7 @@
{
percentiles = pool_nrealloc (syntax_pool, percentiles,
n_percentiles + 1, sizeof *percentiles);
-
- if (i < n_percentiles)
- memmove (&percentiles[i + 1], &percentiles[i],
- (n_percentiles - i) * sizeof (struct percentile) );
-
+ insert_element (percentiles, n_percentiles, sizeof *percentiles, i);
percentiles[i].p = x;
n_percentiles++;
}
@@ -1236,28 +1237,9 @@
double rank;
int i = 0;
int idx;
- double *median_value;
/* Calculate percentiles. */
- /* If the 50th percentile was not explicitly requested then we must
- calculate it anyway --- it's the median */
- median_value = 0 ;
- for (i = 0; i < n_percentiles; i++)
- {
- if (percentiles[i].p == 0.5)
- {
- median_value = &percentiles[i].value;
- break;
- }
- }
-
- if ( 0 == median_value )
- {
- add_percentile (0.5);
- implicit_50th = 1;
- }
-
for (i = 0; i < n_percentiles; i++)
{
percentiles[i].flag = 0;
@@ -1337,9 +1319,6 @@
percentiles[i].value = percentiles[i].x1 +
( percentiles[i].x2 - percentiles[i].x1) * s ;
-
- if ( percentiles[i].p == 0.50)
- median_value = &percentiles[i].value;
}
@@ -1376,7 +1355,6 @@
d[frq_max] = ft->valid[ft->n_valid - 1].value[0].f;
d[frq_mode] = X_mode;
d[frq_range] = d[frq_max] - d[frq_min];
- d[frq_median] = *median_value;
d[frq_sum] = d[frq_mean] * W;
d[frq_stddev] = sqrt (d[frq_variance]);
d[frq_semean] = d[frq_stddev] / sqrt (W);
@@ -1393,11 +1371,6 @@
struct tab_table *t;
int i, r;
- int n_explicit_percentiles = n_percentiles;
-
- if ( implicit_50th && n_percentiles > 0 )
- --n_percentiles;
-
if (var_is_alpha (v))
return;
ft = &get_var_freqs (v)->tab;
@@ -1409,7 +1382,7 @@
}
calc_stats (v, stat_value);
- t = tab_create (3, n_stats + n_explicit_percentiles + 2, 0);
+ t = tab_create (3, n_stats + n_percentiles + 2, 0);
tab_dim (t, tab_natural_dimensions);
tab_box (t, TAL_1, TAL_1, -1, -1 , 0 , 0 , 2, tab_nr(t) - 1) ;
@@ -1437,14 +1410,17 @@
tab_float(t, 2, 1, TAB_NONE, ft->total_cases - ft->valid_cases, 11, 0);
- for (i = 0; i < n_explicit_percentiles; i++, r++)
+ for (i = 0; i < n_percentiles; i++, r++)
{
if ( i == 0 )
{
tab_text (t, 0, r, TAB_LEFT | TAT_TITLE, _("Percentiles"));
}
- tab_float (t, 1, r, TAB_LEFT, percentiles[i].p * 100, 3, 0 );
+ if (percentiles[i].p == 0.5)
+ tab_text (t, 1, r, TAB_LEFT, _("50 (Median)"));
+ else
+ tab_float (t, 1, r, TAB_LEFT, percentiles[i].p * 100, 3, 0);
tab_float (t, 2, r, TAB_NONE, percentiles[i].value, 11, 3);
}
Index: tests/ChangeLog
===================================================================
RCS file: /cvsroot/pspp/pspp/tests/ChangeLog,v
retrieving revision 1.100
retrieving revision 1.101
diff -u -b -r1.100 -r1.101
--- tests/ChangeLog 28 Jul 2007 00:40:53 -0000 1.100
+++ tests/ChangeLog 2 Aug 2007 03:07:17 -0000 1.101
@@ -1,3 +1,12 @@
+2007-08-01 Ben Pfaff <address@hidden>
+
+ * command/weight.sh: Update to match new output format for median
+ under FREQUENCIES.
+
+ * stats/percentiles-compatible.sh: Ditto.
+
+ * stats/percentiles-enhanced.sh: Ditto.
+
2007-07-28 John Darrington <address@hidden>
* command/t-test-1-indep-val.sh: Changed the order of groups in the
Index: tests/command/weight.sh
===================================================================
RCS file: /cvsroot/pspp/pspp/tests/command/weight.sh,v
retrieving revision 1.19
retrieving revision 1.20
diff -u -b -r1.19 -r1.20
--- tests/command/weight.sh 5 Dec 2006 15:10:22 -0000 1.19
+++ tests/command/weight.sh 2 Aug 2007 03:07:17 -0000 1.20
@@ -146,12 +146,11 @@
| 80| 1| 0|100|
| 94| 1| 0|100|
+--------+--------+---+---+
-+-----------------+---------+
++-----------------------+---------+
|N Valid | 730|
-| Missing| 0|
+| Missing | 0|
|Mean | 31.515|
|S.E. Mean | .405|
-|Median | 28.500|
|Mode | 21.000|
|Std Dev | 10.937|
|Variance | 119.608|
@@ -163,7 +162,8 @@
|Minimum | 18.000|
|Maximum | 94.000|
|Sum |23006.000|
-+-----------------+---------+
+|Percentiles 50 (Median)| 28.500|
++-----------------------+---------+
EOF
if [ $? -ne 0 ] ; then fail ; fi
Index: tests/stats/percentiles-compatible.sh
===================================================================
RCS file: /cvsroot/pspp/pspp/tests/stats/percentiles-compatible.sh,v
retrieving revision 1.10
retrieving revision 1.11
diff -u -b -r1.10 -r1.11
--- tests/stats/percentiles-compatible.sh 5 Dec 2006 15:10:22 -0000
1.10
+++ tests/stats/percentiles-compatible.sh 2 Aug 2007 03:07:18 -0000
1.11
@@ -98,19 +98,19 @@
#===========#========#=========#========#========#========#
| Total| 5| 100.0| 100.0| |
+--------------------+---------+--------+--------+--------+
-+-------------------+-----+
++-----------------------+-----+
|N Valid | 5|
-| Missing| 0|
+| Missing | 0|
|Mean |3.000|
|Std Dev |1.581|
|Minimum |1.000|
|Maximum |5.000|
|Percentiles 0 |1.000|
| 25 |1.500|
-| 50 |3.000|
+| 50 (Median)|3.000|
| 75 |4.500|
| 100 |5.000|
-+-------------------+-----+
++-----------------------+-----+
EOF
if [ $? -ne 0 ] ; then fail ; fi
Index: tests/stats/percentiles-enhanced.sh
===================================================================
RCS file: /cvsroot/pspp/pspp/tests/stats/percentiles-enhanced.sh,v
retrieving revision 1.10
retrieving revision 1.11
diff -u -b -r1.10 -r1.11
--- tests/stats/percentiles-enhanced.sh 5 Dec 2006 15:10:22 -0000 1.10
+++ tests/stats/percentiles-enhanced.sh 2 Aug 2007 03:07:18 -0000 1.11
@@ -97,19 +97,19 @@
#===========#========#=========#========#========#========#
| Total| 5| 100.0| 100.0| |
+--------------------+---------+--------+--------+--------+
-+-------------------+-----+
++-----------------------+-----+
|N Valid | 5|
-| Missing| 0|
+| Missing | 0|
|Mean |3.000|
|Std Dev |1.581|
|Minimum |1.000|
|Maximum |5.000|
|Percentiles 0 |1.000|
| 25 |2.000|
-| 50 |3.000|
+| 50 (Median)|3.000|
| 75 |4.000|
| 100 |5.000|
-+-------------------+-----+
++-----------------------+-----+
EOF
if [ $? -ne 0 ] ; then fail ; fi
@@ -160,19 +160,19 @@
#===========#========#=========#========#========#========#
| Total| 10| 100.0| 100.0| |
+--------------------+---------+--------+--------+--------+
-+-------------------+-----+
++-----------------------+-----+
|N Valid | 10|
-| Missing| 0|
+| Missing | 0|
|Mean |3.000|
|Std Dev |1.491|
|Minimum |1.000|
|Maximum |5.000|
|Percentiles 0 |1.000|
| 25 |2.000|
-| 50 |3.000|
+| 50 (Median)|3.000|
| 75 |4.000|
| 100 |5.000|
-+-------------------+-----+
++-----------------------+-----+
EOF
if [ $? -ne 0 ] ; then fail ; fi
@@ -220,19 +220,19 @@
#===========#========#=========#========#========#========#
| Total| 6| 100.0| 100.0| |
+--------------------+---------+--------+--------+--------+
-+-------------------+-----+
++-----------------------+-----+
|N Valid | 6|
-| Missing| 0|
+| Missing | 0|
|Mean |3.500|
|Std Dev |1.517|
|Minimum |1.000|
|Maximum |5.000|
|Percentiles 0 |1.000|
| 25 |3.000|
-| 50 |3.500|
+| 50 (Median)|3.500|
| 75 |4.750|
| 100 |5.000|
-+-------------------+-----+
++-----------------------+-----+
EOF
if [ $? -ne 0 ] ; then fail ; fi
@@ -281,19 +281,19 @@
#===========#========#=========#========#========#========#
| Total| 10| 100.0| 100.0| |
+--------------------+---------+--------+--------+--------+
-+-------------------+-----+
++-----------------------+-----+
|N Valid | 6|
-| Missing| 4|
+| Missing | 4|
|Mean |3.500|
|Std Dev |1.517|
|Minimum |1.000|
|Maximum |5.000|
|Percentiles 0 |1.000|
| 25 |3.000|
-| 50 |3.500|
+| 50 (Median)|3.500|
| 75 |4.750|
| 100 |5.000|
-+-------------------+-----+
++-----------------------+-----+
EOF
if [ $? -ne 0 ] ; then fail ; fi
- [Pspp-cvs] pspp src/language/stats/ChangeLog src/language/...,
Ben Pfaff <=