--- loncom/interface/statistics/lonproblemstatistics.pm 2002/08/13 12:21:46 1.20 +++ loncom/interface/statistics/lonproblemstatistics.pm 2004/07/12 19:30:12 1.90 @@ -1,7 +1,6 @@ # The LearningOnline Network with CAPA -# (Publication Handler # -# $Id: lonproblemstatistics.pm,v 1.20 2002/08/13 12:21:46 stredwic Exp $ +# $Id: lonproblemstatistics.pm,v 1.90 2004/07/12 19:30:12 matthew Exp $ # # Copyright Michigan State University Board of Trustees # @@ -26,720 +25,1707 @@ # http://www.lon-capa.org/ # # (Navigate problems for statistical reports -# YEAR=2001 -# 5/5,7/9,7/25/1,8/11,9/13,9/26,10/5,10/9,10/22,10/26 Behrouz Minaei -# 11/1,11/4,11/16,12/14,12/16,12/18,12/20,12/31 Behrouz Minaei -# YEAR=2002 -# 1/22,2/1,2/6,2/25,3/2,3/6,3/17,3/21,3/22,3/26,4/7,5/6 Behrouz Minaei -# 5/12,5/14,5/15,5/19,5/26,7/16,7/25,7/29,8/5 Behrouz Minaei # -### +############################################### +############################################### -package Apache::lonproblemstatistics; +=pod + +=head1 NAME + +lonproblemstatistics + +=head1 SYNOPSIS + +Routines to present problem statistics to instructors via tables, +Excel files, and plots. + +=over 4 + +=cut + +############################################### +############################################### + +package Apache::lonproblemstatistics; use strict; use Apache::lonnet(); +use Apache::loncommon(); use Apache::lonhtmlcommon; use Apache::loncoursedata; -use GDBM_File; +use Apache::lonstatistics; +use LONCAPA::lonmetadata(); +use Apache::lonlocal; +use Spreadsheet::WriteExcel; +use Apache::lonstathelpers(); +use Time::HiRes; + +my @StatsArray; +my %SeqStat; # keys are symbs, values are hash refs + +## +## Localization notes: +## +## in @Fields[0]->{'long_title'} is placed in Excel files and is used as the +## 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? +## +## format no sprintf format string +## +## excel_format no excel format type +## (see &Apache::loncommon::define_excel_formats +my @Fields = ( + { name => 'problem_num', + title => 'P#', + align => 'right', + color => '#FFFFE6', + selectable => 'no', + defaultselected => 'yes', + }, + { name => 'container', + title => 'Sequence or Folder', + align => 'left', + color => '#FFFFE6', + sortable => 'yes', + selectable => 'no', + defaultselected => 'yes', + }, + { name => 'title', + title => 'Title', + align => 'left', + color => '#FFFFE6', + special => 'link', + sortable => 'yes', + selectable => 'no', + defaultselected => 'yes', + }, + { name => 'part', + title => 'Part', + align => 'left', + color => '#FFFFE6', + selectable => 'no', + defaultselected => 'yes', + }, + { name => 'num_students', + title => '#Stdnts', + align => 'right', + color => '#EEFFCC', + format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Number of Students Attempting Problem', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'tries', + title => 'Tries', + align => 'right', + color => '#EEFFCC', + format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Total Number of Tries', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'max_tries', + title => 'Max Tries', + align => 'right', + color => '#DDFFFF', + format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Maximum Number of Tries', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'min_tries', + title => 'Min Tries', + align => 'right', + color => '#DDFFFF', + format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Minumum Number of Tries', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'mean_tries', + title => 'Mean Tries', + align => 'right', + color => '#DDFFFF', + format => '%5.2f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Average Number of Tries', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'std_tries', + title => 'S.D. tries', + align => 'right', + color => '#DDFFFF', + format => '%5.2f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Standard Deviation of Number of Tries', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'skew_tries', + title => 'Skew Tries', + align => 'right', + color => '#DDFFFF', + format => '%5.2f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Skew of Number of Tries', + selectable => 'yes', + defaultselected => 'no', + }, + { name => 'num_solved', + title => '#YES', + align => 'right', + color => '#FFDDDD', + format => '%4.1f',# format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Number of Students able to Solve', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'num_override', + title => '#yes', + align => 'right', + color => '#FFDDDD', + format => '%4.1f',# format => '%d', + sortable => 'yes', + graphable => 'yes', + long_title => 'Number of Students given Override', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'num_wrong', + title => '#Wrng', + align => 'right', + color => '#FFDDDD', + format => '%4.1f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Percent of students whose final answer is wrong', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'deg_of_diff', + title => 'DoDiff', + align => 'right', + color => '#FFFFE6', + format => '%5.2f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Degree of Difficulty'. + '[ 1 - ((#YES+#yes) / Tries) ]', + selectable => 'yes', + defaultselected => 'yes', + }, + { name => 'deg_of_disc', + title => 'DoDisc', + align => 'right', + color => '#FFFFE6', + format => '%4.2f', + sortable => 'yes', + graphable => 'yes', + long_title => 'Degree of Discrimination', + selectable => 'yes', + defaultselected => 'yes', + }, +## duedate included for research purposes. Commented out most of the time. +# { name => 'duedate', +# title => 'Due Date', +# align => 'left', +# color => '#FFFFFF', +# sortable => 'yes', +# graphable => 'no', +# long_title => 'Due date of resource for instructor', +# selectable => 'no', +# defaultselected => 'yes', +# }, +## opendate included for research purposes. Commented out most of the time. +# { name => 'opendate', +# title => 'Open Date', +# align => 'left', +# color => '#FFFFFF', +# sortable => 'yes', +# graphable => 'no', +# long_title => 'date resource became answerable', +# selectable => 'no', +# defaultselected => 'yes', +# }, +## symb included for research purposes. Commented out most of the time. +# { name => 'symb', +# title => 'Symb', +# align => 'left', +# color => '#FFFFFF', +# sortable => 'yes', +# graphable => 'no', +# long_title => 'Unique LON-CAPA identifier for problem', +# selectable => 'no', +# defaultselected => 'yes', +# }, +## resptypes included for research purposes. Commented out most of the time. +# { name => 'resptypes', +# title => 'Response Types', +# align => 'left', +# color => '#FFFFFF', +# sortable => 'no', +# graphable => 'no', +# long_title => 'Response Types used in this problem', +# selectable => 'no', +# defaultselected => 'yes', +# }, +); + +my @SeqFields = ( + { name => 'title', + title => 'Sequence', + align => 'left', + color => '#FFFFE6', + special => 'no', + sortable => 'no', + selectable => 'yes', + defaultselected => 'no', + }, + { name => 'items', + title => '#Items', + align => 'right', + color => '#FFFFE6', + format => '%4d', + sortable => 'no', + graphable => 'no', + long_title => 'Number of Items in Sequence', + selectable => 'yes', + defaultselected => 'no', + }, + { name => 'scoremean', + title => 'Score Mean', + align => 'right', + color => '#FFFFE6', + format => '%4.2f', + sortable => 'no', + graphable => 'no', + long_title => 'Mean Sequence Score', + selectable => 'yes', + defaultselected => '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', + defaultselected => 'no', + }, + { name => 'scoremax', + title => 'Score Max', + align => 'right', + color => '#FFFFE6', + format => '%4.2f', + sortable => 'no', + graphable => 'no', + long_title => 'Maximum Sequence Score', + selectable => 'yes', + defaultselected => 'no', + }, + { name => 'scoremin', + title => 'Score Min', + align => 'right', + color => '#FFFFE6', + format => '%4.2f', + sortable => 'no', + graphable => 'no', + long_title => 'Minumum Sequence Score', + selectable => 'yes', + defaultselected => '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', + defaultselected => 'no', + }, + { name => 'countmean', + title => 'Count Mean', + align => 'right', + color => '#FFFFFF', + format => '%4.2f', + sortable => 'no', + graphable => 'no', + long_title => 'Mean Sequence Score', + selectable => 'yes', + defaultselected => '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', + defaultselected => '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', + defaultselected => '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', + defaultselected => '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', + defaultselected => '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', + defaultselected => 'no', + }, +); + +my %SelectedFields; + +sub parse_field_selection { + # + # Pull out the defaults + if (! defined($ENV{'form.fieldselections'})) { + $ENV{'form.fieldselections'} = []; + foreach my $field (@Fields) { + next if ($field->{'selectable'} ne 'yes'); + if ($field->{'defaultselected'} eq 'yes') { + push(@{$ENV{'form.fieldselections'}},$field->{'name'}); + } + } + } + # + # Make sure the data we are plotting is there + my %NeededFields; + if (exists($ENV{'form.plot'}) && $ENV{'form.plot'} ne '' && + $ENV{'form.plot'} ne 'none') { + if ($ENV{'form.plot'} eq 'degrees') { + $NeededFields{'deg_of_diff'}++; + $NeededFields{'deg_of_disc'}++; + } elsif ($ENV{'form.plot'} eq 'tries statistics') { + $NeededFields{'mean_tries'}++; + $NeededFields{'std_tries'}++; + $NeededFields{'problem_num'}++; + } else { + $NeededFields{$ENV{'form.plot'}}++; + } + } + # + # This should not happen, but in case it does... + if (ref($ENV{'form.fieldselections'}) ne 'ARRAY') { + $ENV{'form.fieldselections'} = [$ENV{'form.fieldselections'}]; + } + # + # Set the field data and the selected fields (for easier checking) + undef(%SelectedFields); + foreach my $field (@Fields) { + if ($field->{'selectable'} ne 'yes') { + $field->{'selected'} = 'yes'; + } else { + $field->{'selected'} = 'no'; + } + if (exists($NeededFields{$field->{'name'}})) { + $field->{'selected'} = 'yes'; + $SelectedFields{$field->{'name'}}++; + } + foreach my $selection (@{$ENV{'form.fieldselections'}}) { + if ($selection eq $field->{'name'} || $selection eq 'all') { + $field->{'selected'} = 'yes'; + $SelectedFields{$field->{'name'}}++; + } + } + } + # + # Always show all the sequence statistics (for now) + foreach my $field (@SeqFields) { + $field->{'selected'} = 'yes'; + } + return; +} -my $jr; +sub field_selection_input { + my $Str = '\n"; +} -sub BuildProblemStatisticsPage { - my ($cacheDB, $students, $courseID, $c, $r)=@_; - my %cache; +############################################### +############################################### - $jr = $r; +=pod - unless(tie(%cache,'GDBM_File',$cacheDB,&GDBM_READER(),0640)) { - return 'Unable to tie database.'; - } +=item &CreateInterface() - my $Ptr = ''; - $Ptr .= ''; - $Ptr .= ''."\n"; - $Ptr .= ''."\n"; - $Ptr .= ''."\n"; - $Ptr .= ''."\n"; - $Ptr .= &ProblemStatisticsButtons($cache{'DisplayFormat'}, - $cache{'DisplayLegend'}); - $Ptr .= '
Select Map'; - $Ptr .= &Apache::lonhtmlcommon::MapOptions(\%cache, 'ProblemStatistics', - 'Statistics'); - $Ptr .= '
Sorting Type:'."\n"; - $Ptr .= &Apache::lonhtmlcommon::AscendOrderOptions( - $cache{'ProblemStatisticsAscend'}, - 'ProblemStatistics', - 'Statistics'); - $Ptr .= '
'; - if($cache{'DisplayLegend'} eq 'Show Legend') { - $Ptr .= &ProblemStatisticsLegend(); - } - $r->print($Ptr); - $r->rflush(); +Create the main intereface for the statistics page. Allows the user to +select sections, maps, and output. - untie(%cache); +=cut - &Apache::loncoursedata::DownloadStudentCourseDataSeparate($students,'true', - $cacheDB,'true', - 'true',$courseID, - $r, $c); - if($c->aborted()) { return; } +############################################### +############################################### +sub CreateInterface { + my ($r) = @_; + # + &parse_field_selection(); + # + my $Str = ''; + $Str .= &Apache::lonhtmlcommon::breadcrumbs + (undef,'Overall Problem Statistics','Statistics_Overall_Key'); + $Str .= ''."\n"; + $Str .= ''; + $Str .= ''; + $Str .= ''; + $Str .= ''; + $Str .= ''; + $Str .= ''; + $Str .= ''."\n"; + # + $Str .= ''."\n"; + $Str .= '
'.&mt('Sections').''.&mt('Enrollment Status').''.&mt('Sequences and Folders').''.&mt('Statistics').''. + &Apache::lonstathelpers::limit_by_time_form().'
'."\n"; + $Str .= &Apache::lonstatistics::SectionSelect('Section','multiple',5); + $Str .= ''; + $Str .= &Apache::lonhtmlcommon::StatusOptions(undef,undef,5); + $Str .= ''; + # + my $only_seq_with_assessments = sub { + my $s=shift; + if ($s->{'num_assess'} < 1) { + return 0; + } else { + return 1; + } + }; + $Str .= &Apache::lonstatistics::MapSelect('Maps','multiple,all',5, + $only_seq_with_assessments); + $Str .= ''.&field_selection_input(); + $Str .= '
'."\n"; + # + $Str .= '

'.&mt('Status: [_1]', + '' + ). + '

'; + # + $Str .= ''; + $Str .= ' 'x5; + $Str .= 'Plot '.&plot_dropdown().(' 'x10); + # + return $Str; +} - unless(tie(%cache,'GDBM_File',$cacheDB,&GDBM_READER(),0640)) { - return 'Unable to tie database.'; - } +############################################### +############################################### + +=pod - my @Header = ("Homework Sets Order","#Stdnts","Tries","Mod", - "Mean","#YES","#yes","%Wrng","DoDiff", - "S.D.","Skew.","D.F.1st","D.F.2nd","Disc."); - my $color=&setbgcolor(0); +=item &BuildProblemStatisticsPage() -# my %Discuss=&Apache::loncoursedata::LoadDiscussion($courseID); -# my ($upper, $lower) = &Discriminant(\%discriminant,$r); - my ($problemData) = &ExtractStudentData(\%cache, $students); - &CalculateStatistics($problemData); - &SortProblems($problemData, $cache{'ProblemStatisticsSort'}, - $cache{'ProblemStatisticsAscend'}); - #$TempCache= - &BuildStatisticsTable(\%cache, $cache{'DisplayFormat'}, - $problemData, \@Header, $r, $color); - untie(%cache); +Main interface to problem statistics. -# foreach (keys %$TempCache) { -# last if ($c->aborted()); -# if(tie(%cache,'GDBM_File',$cacheDB,&GDBM_WRCREAT(),0640)) { -# $cache{$_}=$TempCache->{$_}; -# untie(%cache); -# } -# } +=cut -# if($c->aborted()) { return; } -# untie(%cache); +############################################### +############################################### +sub BuildProblemStatisticsPage { + my ($r,$c)=@_; + # + my %Saveable_Parameters = ('Status' => 'scalar', + 'statsoutputmode' => 'scalar', + 'Section' => 'array', + 'StudentData' => 'array', + 'Maps' => 'array', + 'fieldselections'=> 'array'); + &Apache::loncommon::store_course_settings('statistics', + \%Saveable_Parameters); + &Apache::loncommon::restore_course_settings('statistics', + \%Saveable_Parameters); + # + &Apache::lonstatistics::PrepareClasslist(); + # + # Clear the package variables + undef(@StatsArray); + undef(%SeqStat); + # + # Finally let the user know we are here + my $interface = &CreateInterface($r); + $r->print($interface); + $r->print(''); + # + my @CacheButtonHTML = + &Apache::lonstathelpers::manage_caches($r,'Statistics','stats_status'); + my $Str; + foreach my $html (@CacheButtonHTML) { + $Str.=$html.(' 'x5); + } + # + $r->print($Str); + if (! exists($ENV{'form.firstrun'})) { + $r->print('

'. + &mt('Press "Generate Statistics" when you are ready.'). + '

'. + &mt('It may take some time to update the student data '. + 'for the first analysis. Future analysis this session '. + ' will not have this delay.'). + '

'); + return; + } + $r->rflush(); + # + # This probably does not need to be done each time we are called, but + # it does not slow things down noticably. + &Apache::loncoursedata::populate_weight_table(); + # + if (exists($ENV{'form.Excel'})) { + &Excel_output($r); + } else { + $r->print(''.' 'x5); + $r->rflush(); + my $count = 0; + foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) { + $count += $seq->{'num_assess_parts'}; + } + if ($count > 10) { + $r->print('

'. + &mt('Compiling statistics for [_1] problems',$count). + '

'); + if ($count > 30) { + $r->print('

'.&mt('This will take some time.').'

'); + } + $r->rflush(); + } + # + my $sortby = $ENV{'form.sortby'}; + $sortby = 'container' if (! defined($sortby) || $sortby =~ /^\s*$/); + my $plot = $ENV{'form.plot'}; + if ($plot eq '' || $plot eq 'none') { + undef($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('

'.&mt('Sequence Statistics'). + &Apache::loncommon::help_open_topic('Statistics_Sequence'). + '

'); + $r->print('
'."\n". + ''."\n". + ''); + $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('
'); + $r->print('
'); + $r->rflush(); return; } -#---- Problem Statistics Web Page --------------------------------------- +########################################################## +########################################################## +## +## HTML output routines +## +########################################################## +########################################################## +sub output_html_by_sequence { + my ($r) = @_; + my $c = $r->connection(); + $r->print(&html_preamble()); + # + foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) { + last if ($c->aborted); + next if ($seq->{'num_assess'} < 1); + $r->print("

".$seq->{'title'}."

". + '
'."\n". + ''."\n". + ''. + &statistics_table_header('no container')."\n"); + my @Data = &compute_statistics_on_sequence($seq); + foreach my $data (@Data) { + $r->print(''.&statistics_html_table_data($data, + 'no container'). + "\n"); + } + $r->print('
'."\n".'
'."\n"); + $r->rflush(); + } + return; +} -sub CreateProblemStatisticsTableHeading { - my ($headings,$r)=@_; +sub output_html_stats { + my ($r)=@_; + &compute_all_statistics($r); + $r->print(&html_preamble()); + &sort_data($ENV{'form.sortby'}); + # + my $count=0; + foreach my $data (@StatsArray) { + if ($count++ % 50 == 0) { + $r->print("\n\n"); + $r->print('
'."\n". + ''."\n". + ''. + ''. + &statistics_table_header(). + "\n"); + } + $r->print(''.&statistics_html_table_data($data)."\n"); + } + $r->print("
\n
\n"); + return; +} +sub html_preamble { my $Str=''; - $Str .= ''."\n"; - $Str .= 'P#'."\n"; - foreach(@$headings) { - $Str .= ''.''.$_.' '."\n"; + $Str .= "

". + $ENV{'course.'.$ENV{'request.course.id'}.'.description'}. + "

\n"; + my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits(); + if (defined($starttime) || defined($endtime)) { + # Inform the user what the time limits on the data are. + $Str .= '

'.&mt('Statistics on submissions from [_1] to [_2]', + &Apache::lonlocal::locallocaltime($starttime), + &Apache::lonlocal::locallocaltime($endtime) + ).'

'; } - $Str .= "\n".''."\n"; - + $Str .= "

".&mt('Compiled on [_1]', + &Apache::lonlocal::locallocaltime(time))."

"; return $Str; } -sub BuildStatisticsTable { - my ($cache,$displayFormat,$data,$headings,$r,$color)=@_; -#6666666 -# my $file="/home/httpd/perl/tmp/183d.txt"; -# open(OUT, ">$file"); -#6666666 -## &Apache::lonstatistics::Create_PrgWin($r); -##777777 -## my (%Activity) = &LoadActivityLog(); -## $r->print(''); -## my ($doDiffFile) = &LoadDoDiffFile(); - -##777777 -## $Str .= &Classify($discriminantFactor, $students); - - my %TempCache; - my $problems = $data->{'problemList'}; - if($displayFormat ne 'Display CSV Format') { - $r->print('
'."\n"); - $r->print(''."\n"); - $r->print(&CreateProblemStatisticsTableHeading($headings, $r)); - } else { - $r->print('
'); +############################################### +############################################### +## +## Misc HTML output routines +## +############################################### +############################################### +sub statistics_html_table_data { + my ($data,$options) = @_; + my $row = ''; + foreach my $field (@Fields) { + next if ($options =~ /no $field->{'name'}/); + next if ($field->{'selected'} ne 'yes'); + $row .= ''; } - - my $count = 1; - foreach(@$problems) { - my ($sequence,$problem,$part)=split(':', $_); -# my $problemRef = ''.$cache->{$problem.':title'}.''; - - my $ref = $cache->{$problem.':title'}; - my $title = $cache->{$problem.':title'}; - my $source = 'source'; - my $tableData = join('&', $ref, $title, $source, - $data->{$_.':studentCount'}, - $data->{$_.':totalTries'}, - $data->{$_.':maxTries'}, - sprintf("%.2f", $data->{$_.':mean'}), - $data->{$_.':correct'}, - $data->{$_.':correctByOverride'}, - sprintf("%.1f", $data->{$_.':percentWrong'}), - sprintf("%.2f", $data->{$_.':degreeOfDifficulty'}), - sprintf("%.1f", $data->{$_.':standardDeviation'}), - sprintf("%.1f", $data->{$_.':skewness'}), - sprintf("%.2f", $data->{$_.':discriminationFactor1'}), - sprintf("%.2f", $data->{$_.':discriminationFactor2'}), - 0); # 0 is for discussion, need to figure out -# $TempCache{'CacheTable:'.$_}=$join; - -#6666666 -# $r->print('
'.$out.'&'.$DoD); -# print (OUT $out.'@'.$DoD.'&'); -#6666666 - -#check with Gerd -# $urlres=~/^(\w+)\/(\w+)/; -# if ($StdNo) { -# &Apache::lonnet::put('nohist_resevaldata',\%storestats, -# $1,$2); -# } -#-------------------------------- Row of statistical table - &TableRow($displayFormat,$tableData,$count,$r,$color); -# $GraphDat->{'GraphGif:'.($count-1)}=$DoD.':'.$Wrng; - $count++; - } -# $TempCache{'ProblemCount'}=$count; - if($cache->{'DisplayFormat'} ne 'Display CSV Format') { - $r->print('
{'align'})) { + $row .= ' align="'.$field->{'align'}.'"'; + } + $row .= '>'; + if (exists($field->{'special'}) && $field->{'special'} eq 'link') { + $row .= ''; + } + if (exists($field->{'format'})) { + $row .= sprintf($field->{'format'},$data->{$field->{'name'}}); + } else { + $row .= $data->{$field->{'name'}}; + } + if (exists($field->{'special'}) && $field->{'special'} eq 'link') { + $row.= ''; + } + $row .= '
'."\n"); - } - $r->print('
'); -#6666666 -# close( OUT ); -#666666 - return \%TempCache; + return $row; } -=pod -sub CacheStatisticsTable { - my ($state,$cache,$headings,$r,$color)=@_; - my @list = (); - my %TempCache; - my %myHeader = reverse( %$headings ); - my $pos = $myHeader{$state}; - if ($pos > 0) {$pos++;} - my $p_count = $cache->{'ProblemCount'}; - - for ( my $k=0; $k<$p_count;$k++) { - my $key=$cache->{'CacheTable:'.$k}; - my @Temp=split(/\&/,$key); - $list[$k]=$Temp[$pos].'+'.$key; - } - - if ($pos>0) { - @list = sort OrderedSort (@list); - } else { - @list = sort (@list); +sub statistics_table_header { + my ($options) = @_; + my $header_row; + foreach my $field (@Fields) { + next if ($options =~ /no $field->{'name'}/); + next if ($field->{'selected'} ne 'yes'); + $header_row .= ''; + if (exists($field->{'sortable'}) && $field->{'sortable'} eq 'yes') { + $header_row .= '{'name'}."'". + ';document.Statistics.submit();">'; + } + $header_row .= &mt($field->{'title'}); + if ($options =~ /sortable/) { + $header_row.= ''; + } + if ($options !~ /no plots/ && + exists($field->{'graphable'}) && + $field->{'graphable'} eq 'yes') { + $header_row.=' ('; + $header_row .= ''; + $header_row .= &mt('plot').')'; + } + $header_row .= ''; } - my $cIdx=0; + return $header_row; +} - if ( $pos == 0 ) { - foreach my $sequence (split(':', $cache->{'orderedSequences'})) { - if($cache->{'ProblemStatisticsMaps'} ne 'All Maps' && - $cache->{'ProblemStatisticsMaps'} ne $cache->{$sequence.':title'}) { - next; - } - if ($cIdx==$p_count) { - return \%TempCache; - } - $r->print(&CreateProblemStatisticsTableHeading( - $cache->{'DisplayFormat'}, - $cache->{$sequence.':source'}, - $cache->{$sequence.':title'}, - $headings,$r)); - - my ($tar)=split(/\&/,$list[$cIdx]); - $tar=~s/\+//eg; - my ($SqOrd)=split(/\@/,$tar); - $sequence+=100; - while ($SqOrd==$sequence && $cIdx<$p_count) { - my($Pre, $Post) = split(/\+/,$list[$cIdx]); - &TableRow($cache,$Post,$cIdx,$cIdx,$r,$color,\%TempCache); - $cIdx++; - my ($tar)=split(/\&/,$list[$cIdx]); - $tar=~s/\+//eg; - ($SqOrd)=split(/\@/,$tar); - } - &CloseTable($cache,$r); - } - } - else { - $r->print(&CreateProblemStatisticsTableHeading( - $cache->{'DisplayFormat'}, - 'Sorted by: ', - $headings->{$pos-1}, - $headings,$r)); - for ( my $nIndex = 0; $nIndex < $p_count; $nIndex++ ) { - my($Pre, $Post) = split(/\+/,$list[$nIndex]); - &TableRow($cache,$Post,$nIndex,$nIndex,$r,$color,\%TempCache); - } - &CloseTable($cache,$r); +sub sequence_html_header { + my $Str .= ''; + foreach my $field (@SeqFields) { +# next if ($field->{'selected'} ne 'yes'); + $Str .= '{'title'}.''; } - - return \%TempCache; + $Str .= ''; + return $Str; } -=cut -sub TableRow { - my ($displayFormat,$Str,$RealIdx,$r,$color)=@_; - my($ref,$title,$source,$StdNo,$TotalTries,$MxTries,$Avg,$YES,$Override, - $Wrng,$DoD,$SD,$Sk,$_D1,$_D2,$DiscNo,$Prob)=split(/\&/,$Str); - my $Ptr; - if($displayFormat eq 'Display CSV Format') { - $Ptr="\n".'
'. - "\n".'"'.$RealIdx.'",'. - "\n".'"'.$title.'",'. - "\n".'"'.$source.'",'. - "\n".'"'.$StdNo.'",'. - "\n".'"'.$TotalTries.'",'. - "\n".'"'.$MxTries.'",'. - "\n".'"'.$Avg.'",'. - "\n".'"'.$YES.'",'. - "\n".'"'.$Override.'",'. - "\n".'"'.$Wrng.'",'. - "\n".'"'.$DoD.'",'. - "\n".'"'.$SD.'",'. - "\n".'"'.$Sk.'",'. - "\n".'"'.$_D1.'",'. - "\n".'"'.$_D2.'"'. - "\n".'"'.$DiscNo.'"'; - $r->print("\n".$Ptr); +sub sequence_html_output { + my ($seq) = @_; + my $data = $SeqStat{$seq->{'symb'}}; + my $row = ''; + foreach my $field (@SeqFields) { + next if ($field->{'selected'} ne 'yes'); + $row .= '{'align'})) { + $row .= ' align="'.$field->{'align'}.'"'; + } + $row .= '>'; + if (exists($field->{'format'})) { + $row .= sprintf($field->{'format'},$data->{$field->{'name'}}); + } else { + $row .= $data->{$field->{'name'}}; + } + $row .= ''; + } + $row .= ''."\n"; + return $row; +} + +#################################################### +#################################################### +## +## Plotting Routines +## +#################################################### +#################################################### +sub make_plot { + my ($r,$plot) = @_; + &compute_all_statistics($r); + &sort_data($ENV{'form.sortby'}); + if ($plot eq 'degrees') { + °rees_plot($r); + } elsif ($plot eq 'tries statistics') { + &tries_data_plot($r); } else { - $Ptr="\n".''. - "\n".''.$RealIdx.''. - "\n".''.$ref.''. - "\n".' '.$StdNo.''. - "\n".''.$TotalTries.''. - "\n".''.$MxTries.''. - "\n".''.$Avg.''. - "\n".' '.$YES.''. - "\n".' '.$Override.''. - "\n".' '.$Wrng.''. - "\n".' '.$DoD.''. - "\n".' '.$SD.''. - "\n".' '.$Sk.''. - "\n".' '.$_D1.''. - "\n".' '.$_D2.''. - "\n".' '.$DiscNo.''; - $r->print("\n".$Ptr.'' ); + &make_single_stat_plot($r,$plot); } + return; +} +sub make_single_stat_plot { + my ($r,$datafield) = @_; + # + my $title; my $yaxis; + foreach my $field (@Fields) { + next if ($field->{'name'} ne $datafield); + $title = $field->{'long_title'}; + $yaxis = $field->{'title'}; + last; + } + if ($title eq '' || $yaxis eq '') { + # datafield is something we do not know enough about to plot + $r->print('

'. + &mt('Unable to plot the requested statistic.'). + '

'); + return; + } + # + # Build up the data sets to plot + my @Labels; + my @Data; + my $max = 1; + foreach my $data (@StatsArray) { + push(@Labels,$data->{'problem_num'}); + push(@Data,$data->{$datafield}); + if ($data->{$datafield}>$max) { + $max = $data->{$datafield}; + } + } + foreach (1,2,3,4,5,10,15,20,25,40,50,75,100,150,200,250,300,500,600,750, + 1000,1500,2000,2500,3000,3500,4000,5000,7500,10000,15000,20000) { + if ($max <= $_) { + $max = $_; + last; + } + } + if ($max > 20000) { + $max = 10000*(int($max/10000)+1); + } + # + $r->print("

".&Apache::loncommon::DrawBarGraph($title, + 'Problem Number', + $yaxis, + $max, + undef, # colors + \@Labels, + \@Data)."

\n"); return; } -# For loading the colored table for display or un-colored for print -sub setbgcolor { - my $PrintTable=shift; - my %color; - if ($PrintTable){ - $color{"gb"}="#FFFFFF"; - $color{"red"}="#FFFFFF"; - $color{"yellow"}="#FFFFFF"; - $color{"green"}="#FFFFFF"; - $color{"purple"}="#FFFFFF"; - } else { - $color{"gb"}="#DDFFFF"; - $color{"red"}="#FFDDDD"; - $color{"yellow"}="#EEFFCC"; - $color{"green"}="#DDFFDD"; - $color{"purple"}="#FFDDFF"; - } - - return \%color; -} - -sub ProblemStatisticsButtons { - my ($displayFormat, $displayLegend)=@_; - - my $Ptr = ''; - $Ptr .= '{'problem_num'}); + my $disc = $data->{'deg_of_disc'}; + my $diff = $data->{'deg_of_diff'}; + push(@Disc,$disc); + push(@Diff,$diff); + # + $ymin = $disc if ($ymin > $disc); + $ymin = $diff if ($ymin > $diff); + $ymax = $disc if ($ymax < $disc); + $ymax = $diff if ($ymax < $diff); + } + # + # Make sure we show relevant information. + if ($ymin < 0) { + if (abs($ymin) < 0.05) { + $ymin = 0; + } else { + $ymin = -1; + } } - $Ptr .= '   '; - $Ptr .= ' 0) { + if (abs($ymax) < 0.05) { + $ymax = 0; + } else { + $ymax = 1; + } + } + # + my $xmax = $Labels[-1]; + if ($xmax > 50) { + if ($xmax % 10 != 0) { + $xmax = 10 * (int($xmax/10)+1); + } } else { - $Ptr .= 'value="Display CSV Format" />'."\n"; + if ($xmax % 5 != 0) { + $xmax = 5 * (int($xmax/5)+1); + } } - $Ptr .= ''; + # + my $discdata .= ''.join(',',@Labels).''.$/. + ''.join(',',@Disc).''.$/; + # + my $diffdata .= ''.join(',',@Labels).''.$/. + ''.join(',',@Diff).''.$/; + # + my $title = 'Degree of Discrimination\nand Degree of Difficulty'; + if ($xmax > 50) { + $title = 'Degree of Discrimination and Degree of Difficulty'; + } + # + $plot=<<"END"; + + + $title + + Problem Number + + $discdata + + + $diffdata + + +END + my $plotresult = + '

'.&Apache::lonxml::xmlparse($r,'web',$plot).'

'.$/; + $r->print($plotresult); + return; +} - return $Ptr; +sub tries_data_plot { + my ($r)=@_; + my $count = scalar(@StatsArray); + my $width = 50 + 10*$count; + $width = 300 if ($width < 300); + my $height = 300; + my $plot = ''; + my @STD; my @Mean; my @Max; my @Min; + my @Labels; + my $ymax = 5; + foreach my $data (@StatsArray) { + my $max = $data->{'mean_tries'} + $data->{'std_tries'}; + $ymax = $max if ($ymax < $max); + $ymax = $max if ($ymax < $max); + push(@Labels,$data->{'problem_num'}); + push(@STD,$data->{'std_tries'}); + push(@Mean,$data->{'mean_tries'}); + } + # + # Make sure we show relevant information. + my $xmax = $Labels[-1]; + if ($xmax > 50) { + if ($xmax % 10 != 0) { + $xmax = 10 * (int($xmax/10)+1); + } + } else { + if ($xmax % 5 != 0) { + $xmax = 5 * (int($xmax/5)+1); + } + } + $ymax = int($ymax)+1+2; + # + my $std_data .= ''.join(',',@Labels).''.$/. + ''.join(',',@Mean).''.$/; + # + my $std_error_data .= ''.join(',',@Labels).''.$/. + ''.join(',',@Mean).''.$/. + ''.join(',',@STD).''.$/; + # + my $title = 'Mean and S.D. of Tries'; + if ($xmax > 25) { + $title = 'Mean and Standard Deviation of Tries'; + } + # + $plot=<<"END"; + + $title + + Problem Number + Number of Tries + + $std_error_data + + + $std_data + + +END + my $plotresult = + '

'.&Apache::lonxml::xmlparse($r,'web',$plot).'

'.$/; + $r->print($plotresult); + return; } -sub ProblemStatisticsLegend { - my $Ptr = ''; - $Ptr = ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= ''; - $Ptr .= '
'; - $Ptr .= '#StdntsTotal number of students attempted the problem.'; - $Ptr .= '
'; - $Ptr .= 'TriesTotal number of tries for solving the problem.'; - $Ptr .= '
'; - $Ptr .= 'ModLargest number of tries for solving the problem by a student.'; - $Ptr .= '
'; - $Ptr .= 'MeanAverage number of tries. [ Tries / #Stdnts ]'; - $Ptr .= '
'; - $Ptr .= '#YESNumber of students solved the problem correctly.'; - $Ptr .= '
'; - $Ptr .= '#yesNumber of students solved the problem by override.'; - $Ptr .= '
'; - $Ptr .= '%WrongPercentage of students who tried to solve the problem '; - $Ptr .= 'but is still incorrect. [ 100*((#Stdnts-(#YES+#yes))/#Stdnts) ]'; - $Ptr .= '
'; - $Ptr .= 'DoDiffDegree of Difficulty of the problem. '; - $Ptr .= '[ 1 - ((#YES+#yes) / Tries) ]'; - $Ptr .= '
'; - $Ptr .= 'S.D.Standard Deviation of the tries. '; - $Ptr .= '[ sqrt(sum((Xi - Mean)^2)) / (#Stdnts-1) '; - $Ptr .= 'where Xi denotes every student\'s tries ]'; - $Ptr .= '
'; - $Ptr .= 'Skew.Skewness of the students tries.'; - $Ptr .= '[(sqrt( sum((Xi - Mean)^3) / #Stdnts)) / (S.D.^3)]'; - $Ptr .= '
'; - $Ptr .= 'Dis.F.Discrimination Factor: A Standard for evaluating the '; - $Ptr .= 'problem according to a Criterion
'; - $Ptr .= '[Applied Criterion in %27 Upper Students - '; - $Ptr .= 'Applied the same Criterion in %27 Lower Students]
'; - $Ptr .= '1st Criterion for Sorting the Students: '; - $Ptr .= 'Sum of Partial Credit Awarded / Total Number of Tries
'; - $Ptr .= '2nd Criterion for Sorting the Students: '; - $Ptr .= 'Total number of Correct Answers / Total Number of Tries'; - $Ptr .= '
Disc.Number of Students had at least one discussion.'; - $Ptr .= '
'; - - return $Ptr; -} - -#------- Processing upperlist and lowerlist according to each problem - -sub ExtractStudentData { - my ($cache, $students)=@_; - -#$Apache::lonxml::debug=1; -#&Apache::lonhomework::showhash(%$cache); -#$Apache::lonxml::debug=0; - - my @problemList=(); - my %problemData; - foreach my $sequence (split(':', $cache->{'orderedSequences'})) { - if($cache->{'ProblemStatisticsMaps'} ne 'All Maps' && - $cache->{'ProblemStatisticsMaps'} ne $cache->{$sequence.':title'}) { +sub plot_dropdown { + my $current = ''; + # + if (defined($ENV{'form.plot'})) { + $current = $ENV{'form.plot'}; + } + # + my @Additional_Plots = ( + { graphable=>'yes', + name => 'degrees', + title => 'Difficulty Indexes' }, + { graphable=>'yes', + name => 'tries statistics', + title => 'Tries Statistics' }); + # + my $Str= "\n".''."\n"; + return $Str; +} - foreach my $problemID (split(':', $cache->{$sequence.':problems'})) { - foreach my $part (split(/\:/,$cache->{$sequence.':'. - $problemID. - ':parts'})) { - my $id = $sequence.':'.$problemID.':'.$part; - push(@problemList, $id); - my $totalTries = 0; - my $totalAwarded = 0; - my $correct = 0; - my $correctByOverride = 0; - my $studentCount = 0; - my $maxTries = 0; - my $totalFirst = 0; - my @studentTries=(); - foreach(@$students) { - my $code = $cache->{"$_:$problemID:$part:code"}; - - if(defined($cache->{$_.':error'}) || $code eq ' ' || - $cache->{"$_:$problemID:NoVersion"} eq 'true') { - next; - } +############################################### +############################################### +## +## Excel output routines +## +############################################### +############################################### +sub Excel_output { + my ($r) = @_; + $r->print('

'.&mt('Preparing Excel Spreadsheet').'

'); + ## + ## Compute the statistics + &compute_all_statistics($r); + my $c = $r->connection; + return if ($c->aborted()); + ## + ## Create the excel workbook + my $filename = '/prtspool/'. + $ENV{'user.name'}.'_'.$ENV{'user.domain'}.'_'. + time.'_'.rand(1000000000).'.xls'; + my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits(); + # + # Create sheet + my $excel_workbook = Spreadsheet::WriteExcel->new('/home/httpd'.$filename); + # + # Check for errors + if (! defined($excel_workbook)) { + $r->log_error("Error creating excel spreadsheet $filename: $!"); + $r->print(&mt("Problems creating new Excel file. ". + "This error has been logged. ". + "Please alert your LON-CAPA administrator.")); + return 0; + } + # + # The excel spreadsheet stores temporary data in files, then put them + # together. If needed we should be able to disable this (memory only). + # The temporary directory must be specified before calling 'addworksheet'. + # File::Temp is used to determine the temporary directory. + $excel_workbook->set_tempdir($Apache::lonnet::tmpdir); + # + # Add a worksheet + my $sheetname = $ENV{'course.'.$ENV{'request.course.id'}.'.description'}; + if (length($sheetname) > 31) { + $sheetname = substr($sheetname,0,31); + } + my $excel_sheet = $excel_workbook->addworksheet( + &Apache::loncommon::clean_excel_name($sheetname)); + # + my $format = &Apache::loncommon::define_excel_formats($excel_workbook); + ## + ## Begin creating excel sheet + ## + my ($rows_output,$cols_output) = (0,0); + # + # Put the course description in the header + $excel_sheet->write($rows_output,$cols_output++, + $ENV{'course.'.$ENV{'request.course.id'}.'.description'}, + $format->{'h1'}); + $cols_output += 3; + # + # Put a description of the sections listed + my $sectionstring = ''; + $excel_sheet->write($rows_output,$cols_output++, + &Apache::lonstathelpers::sections_description + (@Apache::lonstatistics::SelectedSections), + $format->{'h3'}); + $cols_output += scalar(@Apache::lonstatistics::SelectedSections); + # + # Time restrictions + my $time_string; + if (defined($starttime)) { + # call localtime but not lonlocal:locallocaltime because excel probably + # cannot handle localized text. Probably. + $time_string .= 'Data collected from '.localtime($time_string); + if (defined($endtime)) { + $time_string .= ' to '.localtime($endtime); + } + $time_string .= '.'; + } elsif (defined($endtime)) { + # See note above about lonlocal:locallocaltime + $time_string .= 'Data collected before '.localtime($endtime).'.'; + } + if (defined($time_string)) { + $excel_sheet->write($rows_output,$cols_output++,$time_string); + $cols_output+= 5; + } + # + # Put the date in there too + $excel_sheet->write($rows_output,$cols_output++, + 'Compiled on '.localtime(time)); + # + $rows_output++; + $cols_output=0; + ## + ## Sequence Statistics + ## + &write_headers($excel_sheet,$format,\$rows_output,\$cols_output, + \@SeqFields); + foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) { + next if ($seq->{'num_assess'} < 1); + my $data = $SeqStat{$seq->{'symb'}}; + $cols_output=0; + foreach my $field (@SeqFields) { + next if ($field->{'selected'} ne 'yes'); + my $fieldformat = undef; + if (exists($field->{'excel_format'})) { + $fieldformat = $format->{$field->{'excel_format'}}; + } + $excel_sheet->write($rows_output,$cols_output++, + $data->{$field->{'name'}},$fieldformat); + } + $rows_output++; + $cols_output=0; + } + ## + ## Resource Statistics + ## + $rows_output++; + $cols_output=0; + &write_headers($excel_sheet,$format,\$rows_output,\$cols_output, + \@Fields); + # + foreach my $data (@StatsArray) { + $cols_output=0; + foreach my $field (@Fields) { + next if ($field->{'selected'} ne 'yes'); + next if ($field->{'name'} eq 'problem_num'); + my $fieldformat = undef; + if (exists($field->{'excel_format'})) { + $fieldformat = $format->{$field->{'excel_format'}}; + } + $excel_sheet->write($rows_output,$cols_output++, + $data->{$field->{'name'}},$fieldformat); + } + $rows_output++; + $cols_output=0; + } + # + $excel_workbook->close(); + # + # Tell the user where to get their excel file + $r->print('
'. + ''. + &mt('Your Excel Spreadsheet').''."\n"); + $r->rflush(); + return; +} - $studentCount++; - my $tries = $cache->{"$_:$problemID:$part:tries"}; - if($maxTries < $tries) { - $maxTries = $tries; - } - $totalTries += $tries; - push(@studentTries, $tries); +## +## &write_headers +## +sub write_headers { + my ($excel_sheet,$format,$rows_output,$cols_output,$Fields) = @_; + ## + ## First the long titles + foreach my $field (@{$Fields}) { + next if ($field->{'name'} eq 'problem_num'); + next if ($field->{'selected'} ne 'yes'); + if (exists($field->{'long_title'})) { + $excel_sheet->write($$rows_output,${$cols_output}, + $field->{'long_title'}, + $format->{'bold'}); + } else { + $excel_sheet->write($$rows_output,${$cols_output},''); + } + ${$cols_output}+= 1; + } + ${$cols_output} =0; + ${$rows_output}+=1; + ## + ## Then the short titles + foreach my $field (@{$Fields}) { + next if ($field->{'selected'} ne 'yes'); + next if ($field->{'name'} eq 'problem_num'); + # Use english for excel as I am not sure how well excel handles + # other character sets.... + $excel_sheet->write($$rows_output,$$cols_output, + $field->{'title'}, + $format->{'bold'}); + $$cols_output+=1; + } + ${$cols_output} =0; + ${$rows_output}+=1; + return; +} - my $awarded = $cache->{"$_:$problemID:$part:awarded"}; - $totalAwarded += $awarded; +################################################## +################################################## +## +## Statistics Gathering and Manipulation Routines +## +################################################## +################################################## +sub compute_statistics_on_sequence { + my ($seq) = @_; + my @Data; + foreach my $res (@{$seq->{'contents'}}) { + next if ($res->{'type'} ne 'assessment'); + foreach my $part (@{$res->{'parts'}}) { + next if ($res->{'partdata'}->{$part}->{'Survey'}); + # + # This is where all the work happens + my $data = &get_statistics($seq,$res,$part,scalar(@StatsArray)+1); + push (@Data,$data); + push (@StatsArray,$data); + } + } + return @Data; +} - if($code eq '*') { - $correct++; - if($tries == 1) { - $totalFirst++; - } - } elsif($code eq '+') { - $correctByOverride++; +sub compute_all_statistics { + my ($r) = @_; + if (@StatsArray > 0) { + # Assume we have already computed the statistics + return; + } + my $c = $r->connection; + foreach my $seq (&Apache::lonstatistics::Sequences_with_Assess()) { + last if ($c->aborted); + next if ($seq->{'num_assess'} < 1); + &compute_sequence_statistics($seq); + &compute_statistics_on_sequence($seq); + } +} + +sub sort_data { + my ($sortkey) = @_; + return if (! @StatsArray); + # + # Sort the data + my $sortby = undef; + foreach my $field (@Fields) { + if ($sortkey eq $field->{'name'}) { + $sortby = $field->{'name'}; + } + } + if (! defined($sortby) || $sortby eq '' || $sortby eq 'problem_num') { + $sortby = 'container'; + } + if ($sortby ne 'container') { + # $sortby is already defined, so we can charge ahead + if ($sortby =~ /^(title|part)$/i) { + # Alpha comparison + @StatsArray = sort { + lc($a->{$sortby}) cmp lc($b->{$sortby}) || + lc($a->{'title'}) cmp lc($b->{'title'}) || + lc($a->{'part'}) cmp lc($b->{'part'}); + } @StatsArray; + } else { + # Numerical comparison + @StatsArray = sort { + my $retvalue = 0; + if ($b->{$sortby} eq 'nan') { + if ($a->{$sortby} ne 'nan') { + $retvalue = -1; + } else { + $retvalue = 0; } } - - $problemData{$id.':sequenceTitle'} = - $cache->{$sequence.':title'}; - $problemData{$id.':studentCount'} = $studentCount; - $problemData{$id.':totalTries'} = $totalTries; - $problemData{$id.':studentTries'} = \@studentTries; - $problemData{$id.':totalAwarded'} = $totalAwarded; - $problemData{$id.':correct'} = $correct; - $problemData{$id.':correctByOverride'} = $correctByOverride; - $problemData{$id.':wrong'} = $studentCount - - ($correct + $correctByOverride); - $problemData{$id.':maxTries'} = $maxTries; - $problemData{$id.':totalFirst'} = $totalFirst; - } + if ($a->{$sortby} eq 'nan') { + if ($b->{$sortby} ne 'nan') { + $retvalue = 1; + } + } + if ($retvalue eq '0') { + $retvalue = $b->{$sortby} <=> $a->{$sortby} || + lc($a->{'title'}) <=> lc($b->{'title'}) || + lc($a->{'part'}) <=> lc($b->{'part'}); + } + $retvalue; + } @StatsArray; } } + # + # Renumber the data set + my $count; + foreach my $data (@StatsArray) { + $data->{'problem_num'} = ++$count; + } + return; +} + +######################################################## +######################################################## + +=pod + +=item &get_statistics() - $problemData{'problemList'} = \@problemList; -# $Discussed=0; -# if($Discuss->{"$name:$problem"}) { -# $TotDiscuss++; -# $Discussed=1; -# } +Wrapper routine from the call to loncoursedata::get_problem_statistics. +Calls lonstathelpers::get_time_limits() to limit the data set by time +and &compute_discrimination_factor - return \%problemData; +Inputs: $sequence, $resource, $part, $problem_num + +Returns: Hash reference with statistics data from +loncoursedata::get_problem_statistics. + +=cut + +######################################################## +######################################################## +sub get_statistics { + my ($sequence,$resource,$part,$problem_num) = @_; + # + my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits(); + my $symb = $resource->{'symb'}; + my $courseid = $ENV{'request.course.id'}; + # + my $data = &Apache::loncoursedata::get_problem_statistics + (\@Apache::lonstatistics::SelectedSections, + $Apache::lonstatistics::enrollment_status, + $symb,$part,$courseid,$starttime,$endtime); + $data->{'symb'} = $symb; + $data->{'part'} = $part; + $data->{'problem_num'} = $problem_num; + $data->{'container'} = $sequence->{'title'}; + $data->{'title'} = $resource->{'title'}; + $data->{'title.link'} = $resource->{'src'}.'?symb='. + &Apache::lonnet::escape($resource->{'symb'}); + # + if ($SelectedFields{'deg_of_disc'}) { + $data->{'deg_of_disc'} = + &compute_discrimination_factor($resource,$part,$sequence); + } + # + # Store in metadata if computations were done for all students + if ($data->{'num_students'} > 1) { + my @Sections = @Apache::lonstatistics::SelectedSections; + my $sections = '"'.join(' ',@Sections).'"'; + $sections =~ s/&+/_/g; # Ensure no special characters + $data->{'sections'}=$sections; + $data->{'course'} = $ENV{'request.course.id'}; + my $urlres=(&Apache::lonnet::decode_symb($resource->{'symb'}))[2]; + $data->{'urlres'}=$urlres; + my %storestats = + &LONCAPA::lonmetadata::dynamic_metadata_storage($data); + my ($dom,$user) = $urlres=~/^(\w+)\/(\w+)/; + &Apache::lonnet::put('nohist_resevaldata',\%storestats,$dom,$user); + } + # + # Get the due date for research purposes (commented out most of the time) +# $data->{'duedate'} = +# &Apache::lonnet::EXT('resource.'.$part.'.duedate',$symb); +# $data->{'opendate'} = +# &Apache::lonnet::EXT('resource.'.$part.'.opendate',$symb); +# $data->{'resptypes'} = join(',',@{$resource->{'partdata'}->{$part}->{'ResponseTypes'}}); + return $data; } -sub SortProblems { - my ($problemData,$sortBy,$ascend)=@_; +############################################### +############################################### - if($sortBy eq "Homework Sets Order") { - return; - } +=pod + +=item &compute_discrimination_factor() + +Inputs: $Resource, $Sequence - my $data; +Returns: integer between -1 and 1 + +=cut - if ($sortBy eq "#Stdnts") { $data = ':studentCount'; } - elsif($sortBy eq "Tries") { $data = ':totalTries'; } - elsif($sortBy eq "Mod") { $data = ':maxTries'; } - elsif($sortBy eq "Mean") { $data = ':mean'; } - elsif($sortBy eq "#YES") { $data = ':correct'; } - elsif($sortBy eq "#yes") { $data = ':correctByOverride'; } - elsif($sortBy eq "%Wrng") { $data = ':percentWrong'; } - elsif($sortBy eq "DoDiff") { $data = ':degreeOfDifficulty'; } - elsif($sortBy eq "S.D.") { $data = ':standardDeviation'; } - elsif($sortBy eq "Skew.") { $data = ':skewness'; } - elsif($sortBy eq "D.F.1st") { $data = ':discriminantFactor1'; } - elsif($sortBy eq "D.F.2nd") { $data = ':discriminantFactor2'; } - elsif($sortBy eq "Disc.") { $data = ''; } - else { return; } - - my $problems = $problemData->{'problemList'}; - my @orderedProblems = - sort { $problemData->{$a.$data} <=> $problemData->{$b.$data} } - @$problems; - if($ascend eq 'Descending') { - @orderedProblems = reverse(@orderedProblems); - } - - $problemData->{'problemList'} = \@orderedProblems; - - return; -} - -sub CalculateStatistics { - my ($data)=@_; - - my $problems = $data->{'problemList'}; - foreach(@$problems) { - # Mean - $data->{$_.':mean'} = ($data->{$_.':studentCount'}) ? - ($data->{$_.':totalTries'} / $data->{$_.':studentCount'}) : 0; - - # %Wrong - $data->{$_.':percentWrong'} = ($data->{$_.':studentCount'}) ? - (($data->{$_.':wrong'} / $data->{$_.':studentCount'}) * 100.0) : - 100.0; - - # Degree of Difficulty - $data->{$_.':degreeOfDifficulty'} = ($data->{$_.':totalTries'}) ? - (1 - (($data->{$_.':correct'} + $data->{$_.':correctByOverride'}) / - $data->{$_.':totalTries'})) : 0; - - # Factor in mean - my $studentTries = $data->{$_.':studentTries'}; - foreach(my $index=0; $index < scalar(@$studentTries); $index++) { - $studentTries->[$index] -= $data->{$_.':mean'}; - } - my $sumSquared = 0; - my $sumCubed = 0; - foreach(@$studentTries) { - my $squared = ($_ * $_); - my $cubed = ($squared * $_); - $sumSquared += $squared; - $sumCubed += $cubed; - } - - # Standard deviation - $data->{$_.':standardDeviation'} = ($data->{$_.':studentCount'} - 1) ? - ((sqrt($sumSquared)) / ($data->{$_.':studentCount'} - 1)) : 0; - - # Skewness - my $standardDeviation = $data->{$_.':standardDeviation'}; - $data->{$_.':skewness'} = ($data->{$_.':standardDeviation'}) ? - (((sqrt($sumSquared)) / $data->{$_.':studentCount'}) / - ($standardDeviation * $standardDeviation * $standardDeviation)) : - 0; - - # Discrimination Factor 1 - $data->{$_.':discriminationFactor1'} = 0; - - # Discrimination Factor 2 - $data->{$_.':discriminationFactor2'} = 0; +############################################### +############################################### +sub compute_discrimination_factor { + my ($resource,$part,$sequence) = @_; + my @Resources; + foreach my $res (@{$sequence->{'contents'}}) { + next if ($res->{'symb'} eq $resource->{'symb'}); + push (@Resources,$res->{'symb'}); + } + # + # rank + my ($starttime,$endtime) = &Apache::lonstathelpers::get_time_limits(); + my $ranking = + &Apache::loncoursedata::rank_students_by_scores_on_resources + (\@Resources, + \@Apache::lonstatistics::SelectedSections, + $Apache::lonstatistics::enrollment_status,undef, + $starttime,$endtime); + # + # compute their percent scores on the problems in the sequence, + my $number_to_grab = int(scalar(@{$ranking})/4); + my $num_students = scalar(@{$ranking}); + my @BottomSet = map { $_->[&Apache::loncoursedata::RNK_student()]; + } @{$ranking}[0..$number_to_grab]; + my @TopSet = + map { + $_->[&Apache::loncoursedata::RNK_student()]; + } @{$ranking}[($num_students-$number_to_grab)..($num_students-1)]; + my ($bottom_sum,$bottom_max) = + &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@BottomSet, + undef,$starttime,$endtime); + my ($top_sum,$top_max) = + &Apache::loncoursedata::get_sum_of_scores($resource,$part,\@TopSet, + undef,$starttime,$endtime); + my $deg_of_disc; + if ($top_max == 0 || $bottom_max==0) { + $deg_of_disc = 'nan'; + } else { + $deg_of_disc = ($top_sum/$top_max) - ($bottom_sum/$bottom_max); } + #&Apache::lonnet::logthis(' '.$top_sum.'/'.$top_max. + # ' - '.$bottom_sum.'/'.$bottom_max); + return $deg_of_disc; +} +############################################### +############################################### +## +## Compute KR-21 +## +## To compute KR-21, you need the following information: +## +## K=the number of items in your test +## M=the mean score on the test +## s=the standard deviation of the scores on your test +## +## then: +## +## KR-21 rk= [K/(K-1)] * [1- (M*(K-M))/(K*s^2))] +## +############################################### +############################################### +sub compute_sequence_statistics { + my ($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}->{'countmean'} = $cMean; + $SeqStat{$symb}->{'count'} = $ccount; + $SeqStat{$symb}->{'items'} = $K; + $SeqStat{$symb}->{'KR-21'}=$kr_21; return; } -sub ProcessDiscriminant { - my ($List) = @_; - my @sortedList = sort (@$List); - my $Count = scalar @sortedList; - my $Problem; - my @Dis; - my $Slvd=0; - my $tmp; - my $Sum1=0; - my $Sum2=0; - my $nIndex=0; - my $nStudent=0; - my %Proc=undef; - while ($nIndex<$Count) { -# $jr->print("
$nIndex) $sortedList[$nIndex]"); - ($Problem,$tmp)=split(/\=/,$sortedList[$nIndex]); - @Dis=split(/\+/,$tmp); - my $Temp = $Problem; - do { - $nIndex++; - $nStudent++; - $Sum1 += $Dis[0]; - $Sum2 += $Dis[1]; - ($Problem,$tmp)=split(/\=/,$sortedList[$nIndex]); - @Dis=split(/\+/,$tmp); - } while ( $Problem eq $Temp && $nIndex < $Count ); - $Proc{$Temp}=($Sum1/$nStudent).':'.($Sum2/$nStudent); -# $jr->print("
$nIndex) $Temp --> ($nStudent) $Proc{$Temp}"); - $Sum1=0; - $Sum2=0; - $nStudent=0; - } - - return %Proc; -} - -#------- Creating Discimination factor -sub Discriminant { - my ($discriminant)=@_; - my @discriminantKeys=keys(%$discriminant); - my $Count = scalar @discriminantKeys; - - my $UpCnt = int(0.27*$Count); - my $low=0; - my $up=$Count-$UpCnt; - my @UpList=(); - my @LowList=(); - - $Count=0; - foreach my $key (sort(@discriminantKeys)) { - $Count++; - if($low < $UpCnt || $Count > $up) { - $low++; - my $str=$discriminant->{$key}; - foreach(split(/\&/,$str)){ - if($_) { - if($low<$UpCnt) { push(@LowList,$_); } - else { push(@UpList,$_); } - } - } - } - } - my %DisUp = &ProcessDiscriminant(\@UpList); - my %DisLow = &ProcessDiscriminant(\@LowList); - return (\%DisUp, \%DisLow); -} -#---- END Problem Statistics Web Page ---------------------------------------- +=pod -#---- Problem Statistics Graph Web Page -------------------------------------- +=item ProblemStatisticsLegend -# ------------------------------------------- Prepare data for Graphical chart +=over 4 -sub BuildGraphicChart { - my ($ylab,$r,$cacheDB)=@_; - my %cache; - my $Col; - my $data=''; - my $count = 0; - my $Max = 0; +=item #Stdnts +Total number of students attempted the problem. - unless(tie(%cache,'GDBM_File',$cacheDB,&GDBM_READER(),0640)) { - return 'Unable to tie database.'; - } - - my $p_count = $cache{'ProblemCount'}; +=item Tries +Total number of tries for solving the problem. - for ( my $k=0; $k<$p_count;$k++) { - my @Temp=split(/\:/,$cache{'GraphGif:'.$k}); - my $inf = $Temp[$Col]; - if ( $Max < $inf ) {$Max = $inf;} - $data .= $inf.','; - $count++; - } - untie(%cache); -# $r->print("
count=$p_count >>data= $data"); +=item Max Tries +Largest number of tries for solving the problem by a student. - if ( $Max > 1 ) { - $Max += (10 - $Max % 10); - $Max = int($Max); - } else { $Max = 1; } +=item Mean +Average number of tries. [ Tries / #Stdnts ] - my $cid=$ENV{'request.course.id'}; +=item #YES +Number of students solved the problem correctly. - if ( $ylab eq 'DoDiff Graph' ) { - $ylab = 'Degree-of-Difficulty'; - $Col = 0; - } else { - $ylab = 'Wrong-Percentage'; - $Col = 1; - } - my $Course = $ENV{'course.'.$cid.'.description'}; - $Course =~ s/\ /"_"/eg; - my $GData=$Course.'&'.'Problems#'.'&'.$ylab.'&'. - $Max.'&'.$count.'&'.$data; +=item #yes +Number of students solved the problem by override. - $r->print(''); +=item %Wrong +Percentage of students who tried to solve the problem +but is still incorrect. [ 100*((#Stdnts-(#YES+#yes))/#Stdnts) ] - return; -} +=item DoDiff +Degree of Difficulty of the problem. +[ 1 - ((#YES+#yes) / Tries) ] + +=item S.D. +Standard Deviation of the tries. +[ sqrt(sum((Xi - Mean)^2)) / (#Stdnts-1) +where Xi denotes every student\'s tries ] + +=item Skew. +Skewness of the students tries. +[(sqrt( sum((Xi - Mean)^3) / #Stdnts)) / (S.D.^3)] + +=item Dis.F. +Discrimination Factor: A Standard for evaluating the +problem according to a Criterion
+ +=item [Criterion to group students into %27 Upper Students - +and %27 Lower Students] +1st Criterion for Sorting the Students: +Sum of Partial Credit Awarded / Total Number of Tries +2nd Criterion for Sorting the Students: +Total number of Correct Answers / Total Number of Tries + +=item Disc. +Number of Students had at least one discussion. + +=back + +=cut + +############################################################ +############################################################ 1; __END__