public override RecordSet Initialize(DataSet Data, Predicate Where, FNodeSet Fields, int Clusters)
{
// Get the min of each field //
AggregateSet set1 = new AggregateSet();
for (int i = 0; i < Fields.Count; i++)
{
set1.Add(new AggregateMin(Fields[i].CloneOfMe()), Fields.Alias(i));
}
// Get the max of each field //
AggregateSet set2 = new AggregateSet();
for (int i = 0; i < Fields.Count; i++)
{
set2.Add(new AggregateMax(Fields[i].CloneOfMe()), Fields.Alias(i));
}
// Render the min and max //
RecordSet rs1 = AggregatePlan.Render(Data, Where, new FNodeSet(), set1);
RecordSet rs2 = AggregatePlan.Render(Data, Where, new FNodeSet(), set2);
// Create the output means table //
RecordSet rs = new RecordSet(Schema.Join(new Schema("key int, count double"), rs1.Columns));
// Fill in the gaps //
for (int i = 0; i < Clusters; i++)
{
if (i == 0)
{
RecordBuilder rb = new RecordBuilder();
rb.Add(0);
rb.Add(0D);
rb.Add(rs1[0]);
rs.Add(rb.ToRecord());
}
else if (i == Clusters - 1)
{
RecordBuilder rb = new RecordBuilder();
rb.Add(Clusters - 1);
rb.Add(0D);
rb.Add(rs2[0]);
rs.Add(rb.ToRecord());
}
else
{
RecordBuilder rb = new RecordBuilder();
rb.Add(i);
rb.Add(0D);
for (int j = 0; j < rs1.Columns.Count; j++)
{
double clus = (double)Clusters;
double jay = (double)j;
rb.Add(rs1[0][j].DOUBLE + (rs2[0][j].DOUBLE - rs1[0][j].DOUBLE) / clus * jay);
}
rs.Add(rb.ToRecord());
}
}
return rs;
}
public RecordSet ToInterim()
{
// Get schema //
Schema s = Schema.Join(this._Maps.Columns, this._Reducers.GetInterimSchema);
// Build the table //
RecordSet rs = new RecordSet(s);
// Load //
foreach (KeyValuePair<Record, CompoundRecord> t in this._cache)
{
Record r = Record.Join(t.Key, t.Value.ToRecord());
rs.Add(r);
}
return rs;
}
private void HandleNullCluster(RecordSet Means)
{
// Increment the fail itterations //
this._zero_fail_itterations++;
// Find which nodes are missing //
List<int> MissingKeys = new List<int>();
for (int i = 0; i < this._count; i++)
{
if (Means.Seek(new Cell(i), 0) == -1)
{
this._zero_fail_counts++;
MissingKeys.Add(i);
}
}
// Add back the missing nodes from the current itteration //
foreach (int x in MissingKeys)
{
RecordBuilder rb = new RecordBuilder();
rb.Add(x);
rb.Add(Record.Subrecord(this._means[x], 1, this._means.Columns.Count - 1));
Means.Add(rb.ToRecord());
}
}
