static ModelTree[] buildLevel1()
{
ModelTree[] ms = new ModelTree[10];
for (int j = 0; j < 10; j++)
{
Model m = new Model();
m.id = Global.id++;
m.len = 1000 * 10;
m.freq = 1000; //leaf node
m.error = 10;
m.type = ModelType.Explicit;
m.values = new double[10];
Random r = new Random();
for (int i = 0; i < 10; i++)
{
m.values[i] = r.Next(100);
}
//children
ms[j] = m.construct();
}
Model seasonal = SetSeasonal();
for (int j = 0; j < 10; j++)
{
ModelTree m = ms[j];
m.seasonal = seasonal;
m.children = buildLevel2();
}
return(ms);
}
public void removeChild(int id_)
{
for (int i = 0; i < childs.Count; i++)
{
ModelTree child = (ModelTree)childs[i];
if (child.id == id_)
{
childs.Remove(child);
}
}
convertchildstochildren();
}
static public void BuildNiceTree()
{
ModelTree t = buildLevel0();
double[] data = new double[t.len];
t.Eval(10000);
t.EvalProb(10000, 12);
t.EvalProb(10000, 1);
for (int i = 0; i < t.len; i++)
{
data[i] = t.EvalProb(i, 1);
}
Serlialize(0, t, data);
}
public ModelTree construct()
{
ModelTree t = new ModelTree();
t.range = new Range(0, len);
t.len = len;
t.values = values;
t.seasonal = seasonal;
t.freq = freq;
t.type = type;
t.error = error;
t.children = null;
t.id = id;
t.ts = ts;
return(t);
}
public ModelTree convert(TimeSeries ts)
{
TimeSeries ts_ = new TimeSeries(ts, this);
Model m = new Model(ts_);
ModelTree t = m.construct();
t.range = this;
t.children = new ModelTree[this.children.Count];
int i = 0;
foreach (Range r in this.children)
{
t.children[i] = r.convert(ts);
i++;
}
return(t);
}
void setModels(int len, double[] errors, int shift, int done)
{
double[] newerrors = null;
if (errors.Length > 1)
{
newerrors = new double[errors.Length - 1];
for (int i = 0; i < errors.Length - 1; i++)
{
newerrors[i] = errors[i];
}
}
int kk = (int)Math.Floor((double)ts.Length / len);
children = new ModelTree[kk];
int x = 0;
for (int i = 0; i < kk; i++)
{
if (i == kk - 1)
{
len = ts.Length - x;
}
int s = x;
double[] u = new double[len];
for (int j = 0; j < len; j++)
{
u[j] = ts.data[x++];
}
TimeSeries t = new TimeSeries(u, ts.freq);
ModelTree m = null;
if (shift == 1)
{
m = new ModelTree(t, newerrors, s);
}
else
{
m = new ModelTree(t, errors, s);
}
if (done == 0)
{
m.BuildTree();
}
children[i] = m;
m.start = s;
}
}
static void Serlialize(int st, ModelTree t, double[] uk)
{
bool append = st > 0;
StreamWriter sw = new StreamWriter(dir + "m.txt", append);
t.ToFile(sw);
sw.Close();
Global.id = 0;
t.Set();
StreamWriter mm = new StreamWriter(dir + "b.txt");
t.SerializeAll();
mm.WriteLine(Global.id);
for (int i = 0; i < Global.id; i++)
{
mm.WriteLine(Global.ht[i]);
}
mm.Close();
}
public static void build(int st, int n)
{
double[] uk = utils.File.ReadData(dir + file, st, n);
TimeSeries ts = new TimeSeries(uk, freq);
BuildModel.ts = ts;
ModelTree t = BuildModel.BTree(errors);
/* ModelTree t= new ModelTree(ts, errors,0);
* t.BuildTree();
* t.Clean();*/
Global.id = 0;
t.Set();
//t.Improvetree();
Global.id = 0;
t.Set();
Serlialize(st, t, uk);
}
static ModelTree buildLevel0()
{
Global.id = 0;
Model m = new Model();
m.id = Global.id++;
m.len = 1000 * 10 * 100;
m.freq = 1000 * 10;
m.error = 100;
m.type = ModelType.Explicit;
m.values = new double[100];
Random r = new Random();
for (int i = 0; i < 10; i++)
{
m.values[i] = r.Next(100);
}
ModelTree t = m.construct();
t.seasonal = SetSeasonal();
t.children = buildLevel1();
return(t);
}
public RTree(ArrayList ranges, TimeSeries ts)
{
ArrayList R_j = (ArrayList)ranges[ranges.Count - 1];
root = (Range)R_j[0];
ArrayList R_i;
for (int i = ranges.Count - 1; i >= 0; i--)
{
R_i = (ArrayList)ranges[i];
foreach (Range r_i in R_i)
{
TimeSeries ts_ = new TimeSeries(ts, r_i);
Model m = new Model(ts_);
m.Clean();
m.Set();
ModelTree t = m.construct();
t.range = r_i;
t.children = null;
r_i.t = t;
}
}
for (int i = ranges.Count - 2; i >= 0; i--)
{
R_i = (ArrayList)ranges[i];
R_j = (ArrayList)ranges[i + 1];
foreach (Range r_i in R_i)
{
foreach (Range r_j in R_j)
{
if (r_j.overlap(r_i))
{
r_j.children.Add(r_i);
}
}
}
}
for (int i = ranges.Count - 2; i >= 0; i--)
{
R_i = (ArrayList)ranges[i];
R_j = (ArrayList)ranges[i + 1];
foreach (Range r_i in R_i)
{
foreach (Range r_j in R_j)
{
if (r_j.overlap(r_i))
{
if (r_j.len > r_i.len)
{
if (r_i.t.error > 1)
{
r_j.t.childs.Add(r_i.t);
r_i.t.parent = r_j.t;
}
}
}
}
}
}
//convert childs to children
for (int i = ranges.Count - 1; i >= 0; i--)
{
R_i = (ArrayList)ranges[i];
foreach (Range r_i in R_i)
{
r_i.t.convertchildstochildren();
}
}
}
void setModels(int len, double[] errors, int shift, int done)
{
double[] newerrors = null;
if (errors.Length > 1)
{
newerrors = new double[errors.Length - 1];
for (int i = 0; i < errors.Length - 1; i++)
{
newerrors[i] = errors[i];
}
}
int kk = (int)Math.Floor((double)ts.Length / len);
children = new ModelTree[kk];
int x = 0;
for (int i = 0; i < kk; i++)
{
if (i == kk - 1) len = ts.Length - x;
int s = x;
double[] u = new double[len];
for (int j = 0; j < len; j++) u[j] = ts.data[x++];
TimeSeries t = new TimeSeries(u, ts.freq);
ModelTree m = null;
if (shift == 1)
m = new ModelTree(t, newerrors, s);
else
m = new ModelTree(t, errors, s);
if (done == 0)
{
m.BuildTree();
}
children[i] = m;
m.start = s;
}
}
public ModelTree construct()
{
ModelTree t = new ModelTree();
t.range = new Range(0, len);
t.len = len;
t.values = values;
t.seasonal = seasonal;
t.freq = freq;
t.type = type;
t.error = error;
t.children = null;
t.id = id;
t.ts = ts;
return t;
}
public int CompareTo(object obj)
{
ModelTree tt = (ModelTree)obj;
return(this.range.CompareTo(tt.range));
}
