public static classificationReportStyleDefinition GetDefault(ExperimentRunNameGroups _groups = null)
{
classificationReportStyleDefinition output = new classificationReportStyleDefinition();
if (_groups == null)
{
_groups = new ExperimentRunNameGroups();
_groups.CheckForDefault();
}
output.layerNeedleByName.Add("Trashold", @"([\d]+)#([\d]*)", "Experiments with document selection controled by trashold");
output.groups = _groups;
output.dataFlags.Add("Render", "LinkText,LinkContent,LINCAPT", "A", "Content from link anchor text");
output.dataFlags.Add("Render", "Tokens,TKN", "TKN", "Tokens extracted from URL");
output.dataFlags.Add("Render", "PageText,PageContent", "B", "Page body text, page description and title tag");
output.dataFlags.Add("Scope", "category,InCategory,Category", "C", "Items in the category");
output.dataFlags.Add("Scope", "page", "P", "Pages");
output.dataFlags.Add("Scope", "Dataset,InDataset", "D", "Items in the complete dataset");
output.dataFlags.Add("Scope", "Link", "L", "Links");
output.dataFlags.Add("Function", "selfCentric", "SC", "Compares items with web site / document set");
output.dataFlags.Add("Function", "Inverse", "I", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "ENT", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "LNG", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "DST", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "Divergence", "DIV", "Promotes diversity, items are at greater distance in vector space");
output.dataFlags.Add("Function", "Convergence", "CON", "Promotes convergence of the items, ones at smaller distance are promoted");
output.dataFlags.Add("Function", "Variance", "VAR", "Promotes variance of the items");
output.dataFlags.Add("Function", "Offset", "OFF", "Measures difference between similarity with true label and average similariy with other labels");
//output.dataColumns.Add("Render", "", "Source of the rendered");
//output.dataColumns.Add("Scope", "", "Scope of analysis, performed by the function");
//output.dataColumns.Add("Function", "", "Function performing the analysus");
//output.dataColumns.Add("WeightModel", "", "Model used for term weighting");
return(output);
}
/// <summary>
/// Constructs the table.
/// </summary>
/// <param name="name">The name.</param>
/// <param name="description">The description.</param>
/// <param name="style">The style.</param>
/// <returns></returns>
public DataTable ConstructTable(String _name, String description, classificationReportTableMode mode = classificationReportTableMode.fullTable)
{
//_name = _name.or(name);
var style = local_style;
DataTable table = new DataTable(_name);
table.SetDescription(description);
DataColumn dc_runGroup = null;
if (mode == classificationReportTableMode.fullTable)
{
dc_runGroup = table.Columns.Add("Group");
dc_runGroup.SetDefaultBackground(Color.LightBlue);
dc_runGroup.SetDesc("Experiment subgroup");
dc_runGroup.SetGroup("ID");
dc_runGroup.SetWidth(10);
}
//, "Run Name", "N", typeof(string), dataPointImportance.normal, "", "Run group");
DataColumn dc_runName = table.Columns.Add("RunName"); //.AddColumn("RunName", "Run Name", "N", typeof(string), dataPointImportance.normal);
dc_runName.SetDefaultBackground(Color.LightSkyBlue);
dc_runName.SetGroup("ID");
dc_runName.SetDesc("Custom identification of experiment setup");
dc_runName.SetWidth(20);
foreach (Int32 size in sizes)
{
DataColumn dc = table.Columns.Add("F" + size, typeof(Double)); //, "F1 measure at " + size + " pages per site", "F1", typeof(double), dataPointImportance.normal, "F5", "F1 at " + size);
dc.SetHeading(style.valueToUse.key + " for " + size);
if (size == maxSize)
{
dc.SetDesc(style.valueToUse.key + " at max size");
dc.SetImportance(dataPointImportance.important);
}
else
{
dc.SetDesc(style.valueToUse.key + " at " + size + " selected");
}
dc.SetFormat(style.ScoreFormat);
dc.SetUnit(size.ToString());
dc.SetLetter("FS_l=" + size.ToString());
dc.SetGroup(style.valueToUse.key + " values");
dc.SetDefaultBackground(Color.LightGray);
dc.SetWidth(10);
}
DataColumn dc_meanDistance = null;
DataColumn dc_meanScore = null;
DataColumn dc_maxValue = null;
if (mode == classificationReportTableMode.fullTable)
{
dc_meanScore = table.Columns.Add("Mean", typeof(Double));
dc_meanDistance = table.Columns.Add("Distance", typeof(Double));
dc_meanDistance.SetDefaultBackground(Color.LightBlue);
dc_meanDistance.SetFormat(style.ScoreFormat);
dc_meanDistance.SetGroup(style.valueToUse.key + " values");
dc_meanDistance.SetWidth(10).SetLetter("∆F_SL").SetDesc("Effectiveness of the selection method (∆FSL), at given selection limit (SL), is expressed as difference between the F1-score (FSL) and the reference score produced without page selection (FREF) method. Mean distance ∆FSL for SL=[1, 10] is adopted as the evaluation criterion.");
dc_meanDistance.SetImportance(dataPointImportance.important);
dc_meanScore.SetDefaultBackground(Color.LightBlue);
dc_meanScore.SetFormat(style.ScoreFormat);
dc_meanScore.SetGroup(style.valueToUse.key + " values").SetWidth(10);
dc_meanScore.SetImportance(dataPointImportance.important);
dc_maxValue = table.Columns.Add("Max", typeof(Double));
dc_maxValue.SetDefaultBackground(Color.LightBlue);
dc_maxValue.SetFormat(style.ScoreFormat);
dc_maxValue.SetGroup(style.valueToUse.key + " values").SetWidth(10);
dc_maxValue.SetLetter("max");
foreach (var dataColumn in style.dataColumns)
{
String ck = "DATA_" + dataColumn.key;
if (!table.Columns.Contains(ck))
{
var cln = table.Columns.Add(ck, typeof(String));
cln.SetHeading(dataColumn.key);
cln.SetGroup("Expanded data");
cln.SetDefaultBackground(Color.Orange);
cln.SetDesc(dataColumn.description);
table.SetAdditionalInfoEntry(dataColumn.key, dataColumn.value, dataColumn.description);
}
}
}
foreach (reportDataFlag flag in style.dataFlags.items)
{
table.SetAdditionalInfoEntry(flag.name, flag.replacement, flag.description);
}
DataColumn dc_timestamp_min = null;
DataColumn dc_timestamp_max = null;
DataColumn dc_comment = null;
if (mode == classificationReportTableMode.fullTable)
{
dc_timestamp_min = table.Columns.Add("TimeMin", typeof(String)); //, "Time stamp", "c", typeof(string), dataPointImportance.normal, "", "Created");
dc_timestamp_min.SetWidth(20);
dc_timestamp_min.SetDesc("Creation time of the oldest report for this runName");
dc_timestamp_max = table.Columns.Add("TimeMax", typeof(String)); //, "Time stamp", "c", typeof(string), dataPointImportance.normal, "", "Created");
dc_timestamp_max.SetWidth(20);
dc_timestamp_max.SetDesc("Creation time of the newest report for this runName");
dc_comment = table.Columns.Add("Comment"); //, "Comment", "i", typeof(string), dataPointImportance.normal);
dc_comment.SetWidth(150);
dc_comment.SetDesc("Report comment, extracted from one of the reports with the same run name");
}
ExperimentRunNameGroups deployedGroups = style.groups.DeployForRunNames(runNames, "FH", "PageContent", "LinkContent");
foreach (var runGroup in deployedGroups.groups)
{
table.SetAdditionalInfoEntry(runGroup.name, runGroup.runNames.toCsvInLine(), runGroup.description);
foreach (var runName in runGroup.runNames)
{
DataRow rw = table.NewRow();
if (dc_runGroup != null)
{
rw[dc_runGroup] = runGroup.name;
}
if (dc_runName != null)
{
rw[dc_runName] = runName;
}
DateTime filecreation_min = DateTime.MaxValue;
DateTime filecreation_max = DateTime.MinValue;
classificationReportExpanded rep = null;
foreach (Int32 size in sizes)
{
String dc_id = "F" + size;
Double val = 0;
if (F1RunNameVsSize[runName].ContainsKey(size))
{
rep = F1RunNameVsSize[runName][size];
val = rep.GetReportValue(style.valueToUse.key);
//val = rep.F1measure;
if (rep.filecreation < filecreation_min)
{
filecreation_min = rep.filecreation;
}
if (rep.filecreation > filecreation_max)
{
filecreation_max = rep.filecreation;
}
}
if (table.Columns.Contains(dc_id))
{
rw[dc_id] = val;
}
}
if (dc_maxValue != null)
{
rw[dc_maxValue] = F1RunNameVsMax[runName];
}
if (dc_meanDistance != null)
{
rw[dc_meanDistance] = F1RunNameVsAverageDistanceFromMax[runName];
}
if (dc_meanScore != null)
{
rw[dc_meanScore] = F1RunNameVsAverage[runName];
}
if (dc_timestamp_min != null)
{
rw[dc_timestamp_min] = filecreation_min.ToString();
}
if (dc_timestamp_max != null)
{
rw[dc_timestamp_max] = filecreation_max.ToString();
}
if (dc_comment != null)
{
rw[dc_comment] = rep.Comment;
}
var dataDict = rep.data.GetDictionary();
foreach (var dataColumn in style.dataColumns)
{
if (dataDict.ContainsKey(dataColumn.key))
{
if (table.Columns.Contains("DATA_" + dataColumn.key))
{
rw["DATA_" + dataColumn.key] = dataDict[dataColumn.key].value;
}
}
}
table.Rows.Add(rw);
}
}
if (dc_maxValue != null)
{
table.GetRowMetaSet().SetStyleForRowsWithValue <Double>(DataRowInReportTypeEnum.dataHighlightA, dc_maxValue.ColumnName, max_total);
}
table.AddExtra("Report space: " + name);
table.AddExtra("Overall maximum: " + max_total.ToString("F5"));
table.AddExtra("Total reports loaded: " + runNames.Count);
table.AddExtra("Group: " + _name);
table.AddExtra("Description: " + description);
table.AddExtra("Dataset: " + dataset);
return(table);
}
/// <summary>
/// Returns clone filtered for matched runNames
/// </summary>
/// <param name="runNames">The run names.</param>
/// <returns></returns>
public ExperimentRunNameGroups DeployForRunNames(IEnumerable <String> runNames, params String[] runNameSuffixes)
{
ExperimentRunNameGroups output = new ExperimentRunNameGroups();
List <String> names = runNames.ToList();
List <String> suffixes = new List <string>();
if (runNameSuffixes != null)
{
suffixes.AddRange(runNameSuffixes);
}
List <String> assigned = new List <string>();
foreach (ExperimentRunNameGroup group in groups)
{
List <String> in_group = new List <string>();
foreach (var in_name in names)
{
if (group.runNames.Contains(in_name))
{
in_group.Add(in_name);
}
else
{
foreach (String suffix in suffixes)
{
if (in_name.EndsWith(suffix))
{
String in_name_nosuffix = in_name.Substring(0, in_name.Length - suffix.Length);
if (group.runNames.Contains(in_name_nosuffix))
{
in_group.Add(in_name);
}
}
}
}
}
if (in_group.Any())
{
assigned.AddRange(in_group);
output.groups.Add(group.DeployForRunNames(in_group));
}
//foreach (var nm in group.runNames)
//{
//}
//List<string> runs = names.Intersect(group.runNames).ToList();
//if (runs.Any())
//{
// assigned.AddRange(runs);
// output.groups.Add(group.DeployForRunNames(runs));
//}
}
foreach (String ass in assigned)
{
names.Remove(ass);
}
if (names.Any())
{
ExperimentRunNameGroup otherGroup = new ExperimentRunNameGroup();
otherGroup.name = "Other";
otherGroup.description = "Ungrouped experiment runs";
otherGroup.runNames.AddRange(names);
otherGroup.Priority = -100;
output.groups.Add(otherGroup);
}
output.groups.OrderByDescending(x => x.Priority);
return(output);
}