public TargetCategoricalClass(SchemaVariable _variable, Node _node)
{
variable = _variable;
node = _node;
ClassSd = new SortedList<string, int>();
Def.Db.TargetCategoricalClassFill(node, ClassSd);
}
public Predictor(SchemaVariable var, Node _node, int predictorLstIdx)
{
ChildrenGroups = new ValueGroup(this, 2);
Variable = var;
node = _node;
PredictorLstIdx = predictorLstIdx;
ValueSd = new SortedList<string,int>();
}
public NodeTargetContinuous(Node ancestor)
: base(ancestor)
{
ImpCalc(); // Outdated use the PreCalculated
//Imp = Ancestor.DescendentImpPreCalculated[ancestor.DescendentLst.Count - 1];
UpdateLevelLst(this);
UI = new NodeTargetContinuousUI(this);
}
public NodeTargetCategorical(Node ancestor)
: base(ancestor)
{
//Fill the TargetCategoricalCases
TargetClasses = new TargetCategoricalClass(Def.Schema.Target, this);
//Imp = ancestor.DescendentImpPreCalculated[ancestor.DescendentLst.Count - 1];
ImpCalc(); // Outdated use the PreCalculated
UpdateLevelLst(this);
if (Def.ExperimentRunning)
return;
UI = new NodeTargetCategoricalUI(this);
}
public Node(Node ancestor)
{
Tree = Def.Tree;
++Node.CountId;
id = Node.CountId;
PositionType = PositionTypeEnum.Intermediate;
Table = new TableMap(this);
Def.Db.PredictorsFill(this);
MvProcedures();
Def.Tree.NodeAdd(this);
Ancestor = ancestor;
Ancestor.DescendentLst.Add(this);
}
//The variation of the index for this predictor ex:
//Sex: female, male offset 1 = male
public PredictorMv(SchemaVariable var, Node _node, int idx, int fieldSpan, int nativeIdx, int offSet, int predictorLstIdx, MvTb MvTb)
{
Variable = var;
N = _node;
PredMvLstIdx = idx;
Table = MvTb;
FieldSpan = fieldSpan;
NativeIdx = nativeIdx;
Offset = offSet;
PredictorLstIdx = predictorLstIdx;
if (Variable.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
LowerNumber = N.PredictorLst[PredictorLstIdx].LowerNumber;
HigherNumber = N.PredictorLst[PredictorLstIdx].HigherNumber;
}
}
public void NodeFill(Node node)
{
if (node.DescendentLst.Count == 0)
return;
Node nl = node.DescendentLst[0], nr = node.DescendentLst[1];
string sql;
double varSum;
Def.Db.ValidationTableIndexDropIfExists(nl.Id);
Def.Db.TableDropIfExists(Def.DbTsTb + nl.Id);
Def.Db.ValidationTableIndexDropIfExists(nr.Id);
Def.Db.TableDropIfExists(Def.DbTsTb + nr.Id);
//Left
sql =
@"CREATE TABLE " +
Def.DbTsTb + nl.Id + "(" + Def.DbTableIdName + " integer NOT NULL)";
Def.Db.ExecuteNonQuery(sql);
//Right node
sql =
@"CREATE TABLE " +
Def.DbTsTb + nr.Id + "(" + Def.DbTableIdName + " integer NOT NULL)";
Def.Db.ExecuteNonQuery(sql);
List<string> insert = new List<string>();
node.MvTsTb.RowCount = (int)Def.Db.GetNumber("Select count(*) from " + Def.DbTsTb + node.Id);
node.MvTsTb.DataFill();
for (int y = 0; y < node.MvTsTb.RowCount; ++y) {
varSum = 0;
for (int i = 0; i < node.PredMvTsLst.Count; ++i) {
varSum += node.PredMvLst[i].Coef * node.PredMvTsLst[i].X(y);
}
if ((varSum + node.C) <= 0) {
insert.Add(@"insert into " + Def.DbTsTb + (nl.Id) + " values (" + node.MvTsTb.Data.ID[y] + ")");
} else {
insert.Add(@"insert into " + Def.DbTsTb + (nr.Id) + " values (" + node.MvTsTb.Data.ID[y] + ")");
}
}
node.MvTb.DataEmpty();
Def.Db.NonQueryTransaction(insert);
Def.Db.ValidationTableIndexCreate(nr.Id);
Def.Db.ValidationTableIndexCreate(nl.Id);
NodeFill(nl);
NodeFill(nr);
}
//Called only by public void RemoveDescendents(Node subtree)
private void SelectNodesToRemove(Node nd, List<Node> toRemove)
{
foreach (Node ndDescendent in nd.DescendentLst) {
toRemove.Add(ndDescendent);
SelectNodesToRemove(ndDescendent, toRemove);
}
nd.DescendentLst.Clear();
return;
}
public bool AutoSplitMv(Node node)
{
int nextNodeId = -1;
string sql = "";
double varSum=0;
NodeTargetCategorical lcat, rcat;
NodeTargetContinuous lcon, rcon;
//if (node.SplitVariable.SplitStatus != Predictor.SplitStatusEnum.CanBeUsed)
// return false;
if (node.Level >= Def.TreeLevelsMax)
return false;
nextNodeId = Node.CountId + 1;
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
//Left node
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId + "(" + Def.DbTableIdName + " integer NOT NULL)";
Def.Db.ExecuteNonQuery(sql);
//Right node
nextNodeId = nextNodeId + 1;
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId + "(" + Def.DbTableIdName + " integer NOT NULL)";
Def.Db.ExecuteNonQuery(sql);
List<string> insert = new List<string>();
for (int y = 0; y < node.Table.RowCount; ++y) {
varSum = 0;
for (int i = 0; i < node.PredMvLst.Count; ++i) {
varSum += node.PredMvLst[i].Coef * node.PredMvLst[i].X(y);
}
if ((varSum + node.C) <= 0) {
insert.Add(@"insert into " + Def.DbTrTb + (nextNodeId - 1) + " values (" + node.MvTb.Data.ID[y] + ")");
} else {
insert.Add(@"insert into " + Def.DbTrTb + (nextNodeId) + " values (" + node.MvTb.Data.ID[y] + ")");
}
}
Def.Db.NonQueryTransaction(insert);
Def.Db.ReferenceTableIndexCreate(nextNodeId-1);
Def.Db.ReferenceTableIndexCreate(nextNodeId);
if (Def.Schema.Target.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
rcon = new NodeTargetContinuous(node);
lcon = new NodeTargetContinuous(node);
// node equation goes here
} else {
rcat = new NodeTargetCategorical(node);
lcat = new NodeTargetCategorical(node);
// node equation goes here
}
afterSplit(node);
return true;
}
public TableMap(Node _node)
{
node = _node;
schema = Def.Schema;
rowCount = (int) Def.Db.GetNumber("Select count(*) from " + Def.DbTrTb + node.Id);
}
public void FullAutogrow(Node node)
{
FrmMessage fmsg = null;
if (!Def.ExperimentRunning) {
Def.FrmMain.WindowState = FormWindowState.Minimized;
fmsg = new FrmMessage("Please, wait", "Calculating...");
fmsg.Show(Def.FrmMain);
Application.DoEvents();
}
FullAutoGrowSubtree(node);
if (!Def.ExperimentRunning) {
Def.FrmMain.WindowState = FormWindowState.Maximized;
fmsg.Close();
Def.FrmMain.TopMost = true;
Application.DoEvents();
Def.FrmMain.TopMost = false;
Def.FrmMain.Focus();
}
Def.Tree.GrowthState = Tree.GrowthStateEnum.FullGrow;
}
public void NodeFill(Node node)
{
if (node.DescendentLst.Count == 0)
return;
Node nl = node.DescendentLst[0], nr = node.DescendentLst[1];
string sql;
int nextNodeId;
Def.Db.TableDropIfExists(Def.DbTsTb + nl.Id);
Def.Db.TableDropIfExists(Def.DbTsTb + nr.Id);
Def.Db.ValidationTableIndexDropIfExists(nl.Id);
Def.Db.ValidationTableIndexDropIfExists(nr.Id);
if (node.SplitVariable.Variable.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
sql =
@"CREATE TABLE " +
Def.DbTsTb + nl.Id +
" AS " +
"SELECT " + Def.DbTsTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTsTb + node.Id + ", " + Def.DbBsTb +
" WHERE " +
Def.DbTsTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
" AND " + node.SplitVariable.Variable.Name + " <= " + node.SplitValue;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ValidationTableIndexCreate(nl.Id);
sql =
@"CREATE TABLE " +
Def.DbTsTb + nr.Id +
" AS " +
"SELECT " + Def.DbTsTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTsTb + node.Id + ", " + Def.DbBsTb +
" WHERE " +
Def.DbTsTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
" AND " + node.SplitVariable.Variable.Name + " > " + node.SplitValue;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ValidationTableIndexCreate(nr.Id);
} else { //IF(nd.SplitVariable.Variable.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Categorical)
string vals = "";
nextNodeId = nl.Id;
//LEFT NODE
List<string> caseLst = node.SplitVariable.ChildrenGroups.ValueGroupLst[0];
for (int i = 0; i < caseLst.Count; ++i) {
vals += node.SplitVariable.Variable.Name + "='" + caseLst[i] + "' ";
if (i < (caseLst.Count - 1)) {
vals += " or ";
} else
vals += ")";
}
sql =
@"CREATE TABLE " +
Def.DbTsTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTsTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTsTb + node.Id + ", " + Def.DbBsTb +
" WHERE (" +
Def.DbTsTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
") AND (" + vals;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ValidationTableIndexCreate(nextNodeId);
//RIGHT NODE
vals = "";
nextNodeId = nr.Id;
caseLst = node.SplitVariable.ChildrenGroups.ValueGroupLst[1];
for (int i = 0; i < caseLst.Count; ++i) {
vals += node.SplitVariable.Variable.Name + "='" + caseLst[i] + "' ";
if (i < (caseLst.Count - 1)) {
vals += " or ";
} else
vals += ")";
}
sql =
@"CREATE TABLE " +
Def.DbTsTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTsTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTsTb + node.Id + ", " + Def.DbBsTb +
" WHERE (" +
Def.DbTsTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
") AND (" + vals;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ValidationTableIndexCreate(nextNodeId);
}
NodeFill(nl);
NodeFill(nr);
}
public void NodeAdd(Node n)
{
NodeLst.Add(n);
return;
}
public void ManuallySplit(Node node)
{
int nextNodeId=-1;
string sql="";
NodeTargetCategorical lcat, rcat;
NodeTargetContinuous lcon, rcon;
if (node.SplitVariable.Variable.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
nextNodeId = Node.CountId+1;
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTrTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTrTb + node.Id + ", " + Def.DbBsTb +
" WHERE " +
Def.DbTrTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
" AND " + node.SplitVariable.Variable.Name + " <= " + node.SplitValue;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ReferenceTableIndexCreate(nextNodeId);
if (Def.Schema.Target.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
lcon = new NodeTargetContinuous(node);
lcon.LabelTopText = " <= " + Math.Round(node.SplitValue, 2);
} else {
lcat = new NodeTargetCategorical(node);
lcat.LabelTopText = " <= " + Math.Round(node.SplitValue, 2);
}
//Right
nextNodeId = Node.CountId + 1;
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTrTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTrTb + node.Id + ", " + Def.DbBsTb +
" WHERE " +
Def.DbTrTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
" AND " + node.SplitVariable.Variable.Name + " > " + node.SplitValue;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ReferenceTableIndexCreate(nextNodeId);
if (Def.Schema.Target.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
rcon = new NodeTargetContinuous(node);
rcon.LabelTopText = " > " + Math.Round(node.SplitValue, 2);
} else {
rcat = new NodeTargetCategorical(node);
rcat.LabelTopText = " > " + Math.Round(node.SplitValue, 2);
}
} else //SchemaVariable.VariableTypeEnum.Continuous
if (node.SplitVariable.Variable.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Categorical) {
string vals = "";
nextNodeId = Node.CountId + 1;
//LEFT NODE
List<string> caseLst = node.SplitVariable.ChildrenGroups.ValueGroupLst[0];
for (int i = 0; i < caseLst.Count; ++i) {
vals += node.SplitVariable.Variable.Name + "='" + caseLst[i] + "' ";
if (i < (caseLst.Count - 1)) {
vals += " or ";
} else
vals += ")";
}
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTrTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTrTb + node.Id + ", " + Def.DbBsTb +
" WHERE (" +
Def.DbTrTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
") AND (" + vals;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ReferenceTableIndexCreate(nextNodeId);
if (Def.Schema.Target.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
lcon = new NodeTargetContinuous(node);
} else {
lcat = new NodeTargetCategorical(node);
}
//RIGHT NODE
vals = "";
nextNodeId = Node.CountId + 1;
caseLst = node.SplitVariable.ChildrenGroups.ValueGroupLst[1];
for (int i = 0; i < caseLst.Count; ++i) {
vals += node.SplitVariable.Variable.Name + "='" + caseLst[i] + "' ";
if (i < (caseLst.Count - 1)) {
vals += " or ";
} else
vals += ")";
}
Def.Db.TableDropIfExists(Def.DbTrTb + nextNodeId);
Def.Db.ReferenceTableIndexDropIfExists(nextNodeId);
sql =
@"CREATE TABLE " +
Def.DbTrTb + nextNodeId +
" AS " +
"SELECT " + Def.DbTrTb + node.Id + "." + Def.DbTableIdName + " FROM " +
Def.DbTrTb + node.Id + ", " + Def.DbBsTb +
" WHERE (" +
Def.DbTrTb + node.Id + "." + Def.DbTableIdName + "=" + Def.DbBsTb + "." + Def.DbTableIdName +
") AND (" + vals;
Def.Db.ExecuteNonQuery(sql);
Def.Db.ReferenceTableIndexCreate(nextNodeId);
if (Def.Schema.Target.VariableTypeUserSet == SchemaVariable.VariableTypeEnum.Continuous) {
rcon = new NodeTargetContinuous(node);
} else {
rcat = new NodeTargetCategorical(node);
}
}
afterSplit(node);
}
public void RemoveDescendents(Node subtree)
{
List<Node> NodeRemoveLst = new List<Node>();
subtree.LabelBottomHide();
SelectNodesToRemove(subtree, NodeRemoveLst);
foreach (Node node in NodeRemoveLst) {
Def.Tree.NodeLst.Remove(node);
node.Dispose();
}
Def.Tree.UpdateLevelsInformation();
subtree.SplitDone = false;
if (subtree is NodeTargetCategorical) {
NodeTargetCategorical nd = (NodeTargetCategorical)subtree;
nd.UI.LabelBottom.Text = "";
} else {
NodeTargetContinuous nd = (NodeTargetContinuous)subtree;
nd.UI.LabelBottom.Text = "";
}
Def.Tree.growthState=GrowthStateEnum.Intermediate;
}
//public abstract void AllGainsCalcMv();
public void UpdateLevelLst(Node n)
{
Level lv;
if (n.PositionType != Node.PositionTypeEnum.Root) {
n.level = n.Ancestor.level + 1;
if (Tree.LevelLast < n.level) {
lv = new Level();
lv.NodeLst.Add(n);
Tree.LevelLst.Add(lv);
} else {
Tree.LevelLst[n.Level].NodeLst.Add(n);
}
}
// LabelBottomText = splitVariable.Variable.Name;
}
public static double VarianceTimesCount(List<string> vals, Node node, Predictor pred)
{
string sql = "", valSql="";
List<double> NLst = null;
for (int i = 0; i < vals.Count; ++i) {
valSql += pred.Variable.Name + "='" + vals[i] + "' ";
if (i < (vals.Count - 1)) {
valSql += " or ";
} else
valSql += ")";
}
sql =
@"SELECT " +
"COALESCE(variance(" + Def.Tree.Schema.Target.Name + "), 0), count(*) " +
"FROM "
+ Def.DbBsTb + " , " + Def.DbTrTb + node.Id + " " +
"WHERE ("
+ Def.DbBsTb + "." + Def.DbTableIdName + "=" +
Def.DbTrTb + node.Id + "." + Def.DbTableIdName + ") and (" +
valSql;
NLst = Def.Db.GetNumberRowLst(sql);
return NLst[0] * NLst[1];
}
private void FullAutoGrowSubtree(Node nd)
{
bool couldSplit = false;
if (nd.Level == Def.TreeLevelsMax)
return;
if (!Def.Multivariate) {
nd.AllGainsCalc();
nd.SearchBestSplit();
nd.SplitVariableIdx = nd.BestSplitIdx;
couldSplit = nd.Tree.AutoSplit(nd);
} else {
nd.AllGainsCalc();
if (nd.SearchBestSplit()) {
nd.SplitVariableIdx = nd.BestSplitIdx;
nd.MvTb.DataFill();
nd.LocalMinimumSearchMv();
couldSplit = nd.Tree.AutoSplitMv(nd);
nd.MvTb.DataEmpty();
}
}
if (couldSplit)
foreach (Node ndDescendent in nd.DescendentLst)
FullAutoGrowSubtree(ndDescendent);
}
private void afterSplit(Node node)
{
// Things to do after any kind of split
node.LabelBottomText = node.SplitVariable.Variable.Name;
node.SplitDone = true;
Def.Tree.GrowthState = Tree.GrowthStateEnum.Intermediate;
CreateLayout();
}