private ClusteredProperties(PivotedProperties pivotedProperties,
IEnumerable <DataPropertyDescriptor> rowHeaders,
Dictionary <string, ClusterRole.Transform> rowTransforms, Dictionary <object, ClusterRole> columnRoles)
{
PivotedProperties = pivotedProperties;
RowHeaders = ImmutableList.ValueOf(rowHeaders);
_rowTransforms = rowTransforms;
_columnRoles = columnRoles;
}
public static Clusterer CreateClusterer(CancellationToken cancellationToken, ClusteringSpec clusteringSpec, ReportResults reportResults)
{
var pivotedPropertySet = new PivotedProperties(reportResults.ItemProperties);
pivotedPropertySet = pivotedPropertySet.ChangeSeriesGroups(pivotedPropertySet.CreateSeriesGroups()).ReorderItemProperties();
var clusteredProperties = ClusteredProperties.FromClusteringSpec(clusteringSpec, pivotedPropertySet);
if (!clusteredProperties.RowValues.Any() && !clusteredProperties.ColumnValues.Any())
{
clusteringSpec = ClusteringSpec.GetDefaultClusteringSpec(cancellationToken, reportResults, pivotedPropertySet);
if (clusteringSpec == null)
{
return(null);
}
clusteredProperties = ClusteredProperties.FromClusteringSpec(clusteringSpec, pivotedPropertySet);
}
return(new Clusterer(cancellationToken, reportResults.RowItems, clusteredProperties, ClusterMetricType.FromName(clusteringSpec.DistanceMetric) ?? ClusterMetricType.DEFAULT));
}
/// <summary>
/// Reorder the ItemProperties collection so that the ungrouped properties come first,
/// followed by the grouped properties.
/// If a group contains multiple series, the properties from those series are interleaved
/// with each other.
/// </summary>
/// <returns></returns>
public PivotedProperties ReorderItemProperties()
{
var groupedPropertyIndexes = SeriesGroups
.SelectMany(group => group.SeriesList.SelectMany(series => series.PropertyIndexes)).ToHashSet();
var newOrder = new List <int>();
newOrder.AddRange(Enumerable.Range(0, ItemProperties.Count).Where(i => !groupedPropertyIndexes.Contains(i)));
newOrder.AddRange(SeriesGroups.SelectMany(group =>
Enumerable.Range(0, group.PivotKeys.Count)
.SelectMany(i => group.SeriesList.Select(series => series.PropertyIndexes[i]))));
var newNumbering = new int[newOrder.Count];
for (int i = 0; i < newOrder.Count; i++)
{
newNumbering[newOrder[i]] = i;
}
var newItemProperties = new ItemProperties(newOrder.Select(i => ItemProperties[i]));
var result = new PivotedProperties(newItemProperties, SeriesGroups.Select(group => group.RenumberProperties(newItemProperties, newNumbering)));
#if DEBUG
Debug.Assert(ItemProperties.ToHashSet().SetEquals(result.ItemProperties.ToHashSet()));
Debug.Assert(SeriesGroups.Count == result.SeriesGroups.Count);
for (int iGroup = 0; iGroup < SeriesGroups.Count; iGroup++)
{
Debug.Assert(SeriesGroups[iGroup].SeriesList.Count == result.SeriesGroups[iGroup].SeriesList.Count);
Debug.Assert(SeriesGroups[iGroup].PivotKeys.SequenceEqual(result.SeriesGroups[iGroup].PivotKeys));
for (int iSeries = 0; iSeries < SeriesGroups[iGroup].SeriesList.Count; iSeries++)
{
var resultSeries = result.SeriesGroups[iGroup].SeriesList[iSeries];
Debug.Assert(resultSeries.PropertyIndexes.OrderBy(i => i).SequenceEqual(resultSeries.PropertyIndexes));
var series = SeriesGroups[iGroup].SeriesList[iSeries];
Debug.Assert(series.PropertyIndexes.Select(i => ItemProperties[i])
.SequenceEqual(resultSeries.PropertyIndexes.Select(i => result.ItemProperties[i])));
}
}
#endif
return(result);
}
public static ClusteringSpec GetDefaultClusteringSpec(CancellationToken cancellationToken, ReportResults reportResults,
PivotedProperties pivotedProperties)
{
var values = new List <ValueSpec>();
foreach (var seriesGroup in pivotedProperties.SeriesGroups)
{
foreach (var series in seriesGroup.SeriesList)
{
var columnRef = ColumnRef.FromPivotedPropertySeries(series);
if (columnRef == null)
{
continue;
}
if (reportResults.RowCount >= MIN_ROWS_TO_ASSUME_HEADER && EqualValuesInAllRows(cancellationToken, reportResults, series))
{
values.Add(new ValueSpec(columnRef, ClusterRole.COLUMNHEADER));
}
else
{
var transform = ClusterRole.IsNumericType(series.PropertyType)
? ClusterRole.ZSCORE
: ClusterRole.BOOLEAN;
values.Add(new ValueSpec(columnRef, transform));
}
}
}
using (var propertyEnumerator = pivotedProperties.UngroupedProperties.GetEnumerator())
{
while (propertyEnumerator.MoveNext())
{
var columnRef = ColumnRef.FromPropertyDescriptor(propertyEnumerator.Current);
if (columnRef == null)
{
continue;
}
values.Insert(0, new ValueSpec(columnRef, ClusterRole.ROWHEADER));
break;
}
if (values.Count == 1)
{
while (propertyEnumerator.MoveNext())
{
var propertyDescriptor = propertyEnumerator.Current;
if (!ClusterRole.IsNumericType(propertyDescriptor?.PropertyType))
{
continue;
}
var columnRef = ColumnRef.FromPropertyDescriptor(propertyDescriptor);
if (columnRef == null)
{
continue;
}
values.Add(new ValueSpec(columnRef, ClusterRole.RAW));
}
if (values.Count == 1)
{
return(null);
}
}
}
if (values.Count == 0)
{
return(null);
}
return(new ClusteringSpec(values).ChangeDistanceMetric(ClusterMetricType.EUCLIDEAN.Name));
}
public static ClusteredProperties FromClusteringSpec(ClusteringSpec clusteringSpec, PivotedProperties pivotedProperties)
{
var allRoles = clusteringSpec.ToValueTransformDictionary();
var rowTransforms = new Dictionary <string, ClusterRole.Transform>();
var rowHeaders = new List <DataPropertyDescriptor>();
foreach (var property in pivotedProperties.UngroupedProperties)
{
var columnRef = ClusteringSpec.ColumnRef.FromPropertyDescriptor(property);
if (columnRef != null && allRoles.TryGetValue(columnRef, out ClusterRole role))
{
if (role == ClusterRole.ROWHEADER)
{
rowHeaders.Add(property);
}
else if (role is ClusterRole.Transform transform)
{
rowTransforms.Add(property.Name, transform);
}
}
}
var columnRoles = new Dictionary <object, ClusterRole>();
foreach (var group in pivotedProperties.SeriesGroups)
{
foreach (var series in group.SeriesList)
{
var columnRef = ToColumnRef(series);
if (columnRef != null && allRoles.TryGetValue(columnRef, out ClusterRole role))
{
if (role == ClusterRole.COLUMNHEADER || role is ClusterRole.Transform)
{
columnRoles.Add(series.SeriesId, role);
}
}
}
}
return(new ClusteredProperties(pivotedProperties, rowHeaders, rowTransforms, columnRoles));
}
public ClusteredProperties ReplacePivotedProperties(PivotedProperties pivotedProperties)
{
return(new ClusteredProperties(pivotedProperties, RowHeaders, _rowTransforms, _columnRoles));
}