[LON-CAPA-cvs] cvs: loncom /interface loncoursedata.pm /interface/statistics lonproblemstatistics.pm
matthew
lon-capa-cvs@mail.lon-capa.org
Thu, 01 Apr 2004 20:02:56 -0000
This is a MIME encoded message
--matthew1080849776
Content-Type: text/plain
matthew Thu Apr 1 15:02:56 2004 EDT
Modified files:
/loncom/interface loncoursedata.pm
/loncom/interface/statistics lonproblemstatistics.pm
Log:
Bug 539: Report KR-21 reliability statistic
loncoursedata.pm:
Added &limit_by_start_end_time to consolidate $starttime and $endtime code.
Added &score_stats and &count_stats to return summary data on the scores
and problem correct counts.
lonproblemstatistics:
Added description of @Fields data structures
Added @SeqFields, &output_sequence_statistics, &sequence_html_header,
&sequence_html_output, and &compute_sequence_statistics.
Reporting statistics on sequences (instead of problems) is still being
fleshed out, interface wise. No Excel output is available for sequence
statistics at this time.
--matthew1080849776
Content-Type: text/plain
Content-Disposition: attachment; filename="matthew-20040401150256.txt"
Index: loncom/interface/loncoursedata.pm
diff -u loncom/interface/loncoursedata.pm:1.128 loncom/interface/loncoursedata.pm:1.129
--- loncom/interface/loncoursedata.pm:1.128 Fri Mar 26 17:01:30 2004
+++ loncom/interface/loncoursedata.pm Thu Apr 1 15:02:55 2004
@@ -1,6 +1,6 @@
# The LearningOnline Network with CAPA
#
-# $Id: loncoursedata.pm,v 1.128 2004/03/26 22:01:30 matthew Exp $
+# $Id: loncoursedata.pm,v 1.129 2004/04/01 20:02:55 matthew Exp $
#
# Copyright Michigan State University Board of Trustees
#
@@ -2160,6 +2160,38 @@
=pod
+=item &limit_by_start_end_times
+
+Build SQL WHERE condition which limits the data collected by the start
+and end times provided
+
+Inputs: $starttime, $endtime, $table
+
+Returns: $time_limits
+
+=cut
+
+##########################################################
+##########################################################
+sub limit_by_start_end_time {
+ my ($starttime,$endtime,$table) = @_;
+ my $time_requirements = undef;
+ if (defined($starttime)) {
+ $time_requirements .= $table.".timestamp>='".$starttime."'";
+ if (defined($endtime)) {
+ $time_requirements .= " AND ".$table.".timestamp<='".$endtime."'";
+ }
+ } elsif (defined($endtime)) {
+ $time_requirements .= $table.".timestamp<='".$endtime."'";
+ }
+ return $time_requirements;
+}
+
+##########################################################
+##########################################################
+
+=pod
+
=item &limit_by_section_and_status
Build SQL WHERE condition which limits the data collected by section and
@@ -2300,6 +2332,147 @@
return ($rows->[0],$rows->[1]);
}
+########################################################
+########################################################
+
+=pod
+
+=item &score_stats
+
+Inputs: $Sections, $enrollment, $symbs, $starttime,
+ $endtime, $courseid
+
+$Sections, $enrollment, $starttime, $endtime, and $courseid are the same as
+elsewhere in this module.
+$symbs is an array ref of symbs
+
+Returns: minimum, maximum, mean, s.d., number of students, and maximum
+ possible of student scores on the given resources
+
+=cut
+
+########################################################
+########################################################
+sub score_stats {
+ my ($Sections,$enrollment,$symbs,$starttime,$endtime,$courseid)=@_;
+ if (! defined($courseid)) {
+ $courseid = $ENV{'request.course.id'};
+ }
+ #
+ &setup_table_names($courseid);
+ my $dbh = &Apache::lonmysql::get_dbh();
+ #
+ my ($section_limits,$enrollment_limits)=
+ &limit_by_section_and_status($Sections,$enrollment,'b');
+ my $time_limits = &limit_by_start_end_time($starttime,$endtime,'a');
+ my @Symbids = map { &get_symb_id($_); } @{$symbs};
+ #
+ my $stats_table = $courseid.'_problem_stats';
+ my $symb_restriction = join(' OR ',map {'a.symb_id='.$_;} @Symbids);
+ my $request = 'DROP TABLE '.$stats_table;
+ $dbh->do($request);
+ $request =
+ 'CREATE TEMPORARY TABLE '.$stats_table.' '.
+ 'SELECT a.student_id,'.
+ 'SUM(a.awarded*w.weight) AS score FROM '.
+ $performance_table.' AS a '.
+ 'NATURAL LEFT JOIN '.$weight_table.' AS w '.
+ 'LEFT JOIN '.$student_table.' AS b ON a.student_id=b.student_id '.
+ 'WHERE ('.$symb_restriction.')';
+ if ($time_limits) {
+ $request .= ' AND '.$time_limits;
+ }
+ if ($section_limits) {
+ $request .= ' AND '.$section_limits;
+ }
+ if ($enrollment_limits) {
+ $request .= ' AND '.$enrollment_limits;
+ }
+ $request .= ' GROUP BY a.student_id';
+# &Apache::lonnet::logthis('request = '.$/.$request);
+ my $sth = $dbh->prepare($request);
+ $sth->execute();
+ $request =
+ 'SELECT AVG(score),STD(score),MAX(score),MIN(score),COUNT(score) '.
+ 'FROM '.$stats_table;
+ my ($ave,$std,$max,$min,$count) = &execute_SQL_request($dbh,$request);
+# &Apache::lonnet::logthis('request = '.$/.$request);
+
+ $request = 'SELECT SUM(weight) FROM '.$weight_table.
+ ' WHERE ('.$symb_restriction.')';
+ my ($max_possible) = &execute_SQL_request($dbh,$request);
+ # &Apache::lonnet::logthis('request = '.$/.$request);
+ return($min,$max,$ave,$std,$count,$max_possible);
+}
+
+
+########################################################
+########################################################
+
+=pod
+
+=item &count_stats
+
+Inputs: $Sections, $enrollment, $symbs, $starttime,
+ $endtime, $courseid
+
+$Sections, $enrollment, $starttime, $endtime, and $courseid are the same as
+elsewhere in this module.
+$symbs is an array ref of symbs
+
+Returns: minimum, maximum, mean, s.d., and number of students
+ of the number of items correct on the given resources
+
+=cut
+
+########################################################
+########################################################
+sub count_stats {
+ my ($Sections,$enrollment,$symbs,$starttime,$endtime,$courseid)=@_;
+ if (! defined($courseid)) {
+ $courseid = $ENV{'request.course.id'};
+ }
+ #
+ &setup_table_names($courseid);
+ my $dbh = &Apache::lonmysql::get_dbh();
+ #
+ my ($section_limits,$enrollment_limits)=
+ &limit_by_section_and_status($Sections,$enrollment,'b');
+ my $time_limits = &limit_by_start_end_time($starttime,$endtime,'a');
+ my @Symbids = map { &get_symb_id($_); } @{$symbs};
+ #
+ my $stats_table = $courseid.'_problem_stats';
+ my $symb_restriction = join(' OR ',map {'a.symb_id='.$_;} @Symbids);
+ my $request = 'DROP TABLE '.$stats_table;
+ $dbh->do($request);
+ $request =
+ 'CREATE TEMPORARY TABLE '.$stats_table.' '.
+ 'SELECT a.student_id,'.
+ 'COUNT(a.award) AS count FROM '.
+ $performance_table.' AS a '.
+ 'LEFT JOIN '.$student_table.' AS b ON a.student_id=b.student_id '.
+ 'WHERE ('.$symb_restriction.')'.
+ " AND a.award!='INCORRECT_ATTEMPTED'";
+ if ($time_limits) {
+ $request .= ' AND '.$time_limits;
+ }
+ if ($section_limits) {
+ $request .= ' AND '.$section_limits;
+ }
+ if ($enrollment_limits) {
+ $request .= ' AND '.$enrollment_limits;
+ }
+ $request .= ' GROUP BY a.student_id';
+ &Apache::lonnet::logthis('request = '.$/.$request);
+ my $sth = $dbh->prepare($request);
+ $sth->execute();
+ $request =
+ 'SELECT AVG(count),STD(count),MAX(count),MIN(count),COUNT(count) '.
+ 'FROM '.$stats_table;
+ my ($ave,$std,$max,$min,$count) = &execute_SQL_request($dbh,$request);
+ &Apache::lonnet::logthis('request = '.$/.$request);
+ return($min,$max,$ave,$std,$count);
+}
######################################################
######################################################
Index: loncom/interface/statistics/lonproblemstatistics.pm
diff -u loncom/interface/statistics/lonproblemstatistics.pm:1.78 loncom/interface/statistics/lonproblemstatistics.pm:1.79
--- loncom/interface/statistics/lonproblemstatistics.pm:1.78 Mon Mar 29 14:50:23 2004
+++ loncom/interface/statistics/lonproblemstatistics.pm Thu Apr 1 15:02:56 2004
@@ -1,6 +1,6 @@
# The LearningOnline Network with CAPA
#
-# $Id: lonproblemstatistics.pm,v 1.78 2004/03/29 19:50:23 matthew Exp $
+# $Id: lonproblemstatistics.pm,v 1.79 2004/04/01 20:02:56 matthew Exp $
#
# Copyright Michigan State University Board of Trustees
#
@@ -61,6 +61,7 @@
use Time::HiRes;
my @StatsArray;
+my %SeqStat; # keys are symbs, values are hash refs
##
## Localization notes:
@@ -69,6 +70,31 @@
## header for plots created with Graph.pm, both of which more than likely do
## not support localization.
##
+#
+#
+##
+## Description of Field attributes
+##
+## Attribute Required Value Meaning or Use
+##
+## name yes any scalar Used to uniquely identify field
+## title yes any scalar This is what the user sees to identify
+## the field. Passed through &mt().
+## long_title yes any scalar Used as graph heading and in excel
+## output. NOT translated
+## align no (left|right|center) HTML cell contents alignment
+## color yes html color HTML cell background color
+## used to visually group statistics
+## special no (link) Indicates a link, target is name.link
+## Currently set in &get_statistics()
+## graphable no (yes|no) Can a bar graph of the field be
+## produced?
+## sortable no (yes|no) Should a sort link be put in the
+## column header?
+## selectable yes (yes|no) Can the column be removed from the
+## statistics display?
+## selected yes (yes|no) Is the column selected by default?
+##
my @Fields = (
{ name => 'problem_num',
title => 'P#',
@@ -236,6 +262,150 @@
},
);
+my @SeqFields = (
+ { name => 'title',
+ title => 'Sequence',
+ align => 'left',
+ color => '#FFFFE6',
+ special => 'no',
+ sortable => 'no',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'items',
+ title => '#Items',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4d',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Number of Items in Sequence',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'scoremean',
+ title => 'Score Mean',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Mean Sequence Score',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'scorestd',
+ title => 'Score STD',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Standard Deviation of Sequence Scores',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'scoremax',
+ title => 'Score Max',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Maximum Sequence Score',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'scoremin',
+ title => 'Score Min',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Minumum Sequence Score',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'scorecount',
+ title => 'Score N',
+ align => 'right',
+ color => '#FFFFE6',
+ format => '%4d',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Number of Students in score computations',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'countmean',
+ title => 'Count Mean',
+ align => 'right',
+ color => '#FFFFFF',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Mean Sequence Score',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'countstd',
+ title => 'Count STD',
+ align => 'right',
+ color => '#FFFFFF',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Standard Deviation of Sequence Scores',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'countmax',
+ title => 'Count Max',
+ align => 'right',
+ color => '#FFFFFF',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Maximum Number of Correct Problems',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'countmin',
+ title => 'Count Min',
+ align => 'right',
+ color => '#FFFFFF',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Minumum Number of Correct Problems',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'count',
+ title => 'Count N',
+ align => 'right',
+ color => '#FFFFFF',
+ format => '%4d',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'Number of Students in score computations',
+ selectable => 'yes',
+ selected => 'no',
+ },
+ { name => 'KR-21',
+ title => 'KR-21',
+ align => 'right',
+ color => '#FFAAAA',
+ format => '%4.2f',
+ sortable => 'no',
+ graphable => 'no',
+ long_title => 'KR-21 reliability statistic',
+ selectable => 'yes',
+ selected => 'no',
+ },
+);
+
my %SelectedFields;
sub parse_field_selection {
@@ -380,6 +550,7 @@
#
# Clear the package variables
undef(@StatsArray);
+ undef(%SeqStat);
#
# Finally let the user know we are here
my $interface = &CreateInterface();
@@ -435,17 +606,40 @@
undef($plot);
}
if ($sortby eq 'container' && ! defined($plot)) {
+ &output_sequence_statistics($r);
&output_html_by_sequence($r);
} else {
if (defined($plot)) {
&make_plot($r,$plot);
}
&output_html_stats($r);
+ &output_sequence_statistics($r);
}
}
return;
}
+sub output_sequence_statistics {
+ my ($r) = @_;
+ my $c=$r->connection();
+ $r->print('<h2>'.&mt('Sequence Statistics').'</h2>');
+ $r->print('<table border="0"><tr><td bgcolor="#777777">'."\n".
+ '<table border="0" cellpadding="3">'."\n".
+ '<tr bgcolor="#FFFFE6">');
+ $r->print(&sequence_html_header());
+ foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) {
+ last if ($c->aborted);
+ next if ($seq->{'num_assess'} < 1);
+ &compute_sequence_statistics($seq);
+ $r->print(&sequence_html_output($seq));
+ }
+ $r->print('</table>');
+ $r->print('</table>');
+ $r->rflush();
+ return;
+}
+
+
##########################################################
##########################################################
##
@@ -584,6 +778,46 @@
return $header_row;
}
+sub sequence_html_header {
+ my $Str .= '<tr>';
+ foreach my $field (@SeqFields) {
+# next if ($field->{'selected'} ne 'yes');
+ $Str .= '<th bgcolor="'.$field->{'color'}.'"';
+ $Str .= '>'.$field->{'title'}.'</th>';
+ }
+ $Str .= '</tr>';
+ return $Str;
+}
+
+
+sub sequence_html_output {
+ my ($seq) = @_;
+ my $data = $SeqStat{$seq->{'symb'}};
+# $SeqStat{$symb}->{'max'}
+# $SeqStat{$symb}->{'min'}
+# $SeqStat{$symb}->{'mean'}
+# $SeqStat{$symb}->{'std'}
+# $SeqStat{$symb}->{'count'}
+# $SeqStat{$symb}->{'max_possible'}
+ my $row = '<tr>';
+ foreach my $field (@SeqFields) {
+# next if ($field->{'selected'} ne 'yes');
+ $row .= '<td bgcolor="'.$field->{'color'}.'"';
+ if (exists($field->{'align'})) {
+ $row .= ' align="'.$field->{'align'}.'"';
+ }
+ $row .= '>';
+ if (exists($field->{'format'})) {
+ $row .= sprintf($field->{'format'},$data->{$field->{'name'}});
+ } else {
+ $row .= $data->{$field->{'name'}};
+ }
+ $row .= '</td>';
+ }
+ $row .= '</tr>'."\n";
+ return $row;
+}
+
####################################################
####################################################
##
@@ -1148,7 +1382,6 @@
return $data;
}
-
###############################################
###############################################
@@ -1205,6 +1438,70 @@
###############################################
###############################################
+##
+## Compute KR-21
+##
+## To compute KR-21, you need the following information:
+##
+## K=the number of items in your test
+## M=the mean score on the test
+## s=the standard deviation of the scores on your test
+##
+## then:
+##
+## KR-21 rk= [K/(K-1)] * [1- (M*(K-M))/(K*s^2))]
+##
+###############################################
+###############################################
+sub compute_sequence_statistics {
+ my ($seq) = @_;
+ my $symb = $seq->{'symb'};
+ my @Resources;
+ foreach my $res (@{$seq->{'contents'}}) {
+ next if ($res->{'type'} ne 'assessment');
+ push (@Resources,$res->{'symb'});
+ }
+ my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits();
+ #
+ # First compute statistics based on student scores
+ my ($smin,$smax,$sMean,$sSTD,$scount,$sMAX) =
+ &Apache::loncoursedata::score_stats
+ (\@Apache::lonstatistics::SelectedSections,
+ $Apache::lonstatistics::enrollment_status,
+ \@Resources,$starttime,$endtime,undef);
+ $SeqStat{$symb}->{'title'} = $seq->{'title'};
+ $SeqStat{$symb}->{'scoremax'} = $smax;
+ $SeqStat{$symb}->{'scoremin'} = $smin;
+ $SeqStat{$symb}->{'scoremean'} = $sMean;
+ $SeqStat{$symb}->{'scorestd'} = $sSTD;
+ $SeqStat{$symb}->{'scorecount'} = $scount;
+ $SeqStat{$symb}->{'max_possible'} = $sMAX;
+ #
+ # Compute statistics based on the number of correct problems
+ # 'correct' is taken to mean
+ my ($cmin,$cmax,$cMean,$cSTD,$ccount)=
+ &Apache::loncoursedata::count_stats
+ (\@Apache::lonstatistics::SelectedSections,
+ $Apache::lonstatistics::enrollment_status,
+ \@Resources,$starttime,$endtime,undef);
+ my $K = $seq->{'num_assess_parts'};
+ my $kr_21;
+ if ($K > 1 && $cSTD > 0) {
+ $kr_21 = ($K/($K-1)) * (1 - $cMean*($K-$cMean)/($K*$cSTD**2));
+ } else {
+ $kr_21 = 'nan';
+ }
+ $SeqStat{$symb}->{'countmax'} = $cmax;
+ $SeqStat{$symb}->{'countmin'} = $cmin;
+ $SeqStat{$symb}->{'countstd'} = $cSTD;
+ $SeqStat{$symb}->{'count'} = $ccount;
+ $SeqStat{$symb}->{'items'} = $K;
+ $SeqStat{$symb}->{'KR-21'}=$kr_21;
+
+ return;
+}
+
+
=pod
@@ -1264,111 +1561,6 @@
=back
=cut
-
-
-############################################################
-############################################################
-##
-## How this all works:
-## Statistics are computed by calling &get_statistics with the sequence,
-## resource, and part id to run statistics on. At various places within
-## the loops which compute the statistics, as well as before and after
-## the entire process, subroutines can be called. The subroutines are
-## registered to the following hooks:
-##
-## hook subroutine inputs
-## ----------------------------------------------------------
-## pre $r,$count
-## pre_seq $r,$count,$seq
-## pre_res $r,$count,$seq,$res
-## calc $r,$count,$seq,$res,$data
-## post_res $r,$count,$seq,$res
-## post_seq $r,$count,$seq
-## post $r,$count
-##
-## abort $r
-##
-## subroutines will be called in the order in which they are registered.
-##
-############################################################
-############################################################
-{
-
-my %hooks;
-my $aborted = 0;
-
-sub abort_computation {
- $aborted = 1;
-}
-
-sub clear_hooks {
- $aborted = 0;
- undef(%hooks);
-}
-
-sub register_hook {
- my ($hookname,$subref)=@_;
- if ($hookname !~ /^(pre|pre_seq|pre_res|post|post_seq|post_res|calc)$/){
- return;
- }
- if (ref($subref) ne 'CODE') {
- &Apache::lonnet::logthis('attempt to register hook to non-code: '.
- $hookname,' = '.$subref);
- } else {
- if (exists($hooks{$hookname})) {
- push(@{$hooks{$hookname}},$subref);
- } else {
- $hooks{$hookname} = [$subref];
- }
- }
- return;
-}
-
-sub run_hooks {
- my $context = shift();
- foreach my $hook (@{$hooks{$context}}) {
- if ($aborted && $context ne 'abort') {
- last;
- }
- my $retvalue = $hook->(@_);
- if (defined($retvalue) && $retvalue eq '0') {
- $aborted = 1 if (! $aborted);
- }
- }
-}
-
-sub run_statistics {
- my ($r) = @_;
- my $count = 0;
- &run_hooks('pre',$r,$count);
- foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) {
- last if ($aborted);
- next if ($seq->{'num_assess'}<1);
- &run_hooks('pre_seq',$r,$count,$seq);
- foreach my $res (@{$seq->{'contents'}}) {
- last if ($aborted);
- next if ($res->{'type'} ne 'assessment');
- &run_hooks('pre_res',$r,$count,$seq,$res);
- foreach my $part (@{$res->{'parts'}}) {
- last if ($aborted);
- #
- # This is where all the work happens
- my $data = &get_statistics($seq,$res,$part,++$count);
- &run_hooks('calc',$r,$count,$seq,$res,$part,$data);
- }
- &run_hooks('post_res',$r,$count,$seq,$res);
- }
- &run_hooks('post_seq',$r,$count,$seq);
- }
- if ($aborted) {
- &run_hooks('abort',$r);
- } else {
- &run_hooks('post',$r,$count);
- }
- return;
-}
-
-} # End of %hooks scope
############################################################
############################################################
--matthew1080849776--