public DataExtremumsForNetwork Clone()
{
var extremums = new DataExtremumsForNetwork();
for (int i = 0; i < InputExtremums.Count; i++)
{
extremums.InputExtremums.Add(InputExtremums[i].Clone());
}
for (int i = 0; i < OutputExtremums.Count; i++)
{
extremums.OutputExtremums.Add(OutputExtremums[i].Clone());
}
return(extremums);
}
public static DataExtremumsForNetwork Merge(DataExtremumsForNetwork x, DataExtremumsForNetwork y)
{
var ret = new DataExtremumsForNetwork();
for (int i = 0; i < x.InputExtremums.Count; i++)
{
ret.InputExtremums.Add(DataExtremum.Merge(x.InputExtremums[i], y.InputExtremums[i]));
}
for (int i = 0; i < x.OutputExtremums.Count; i++)
{
ret.OutputExtremums.Add(DataExtremum.Merge(x.OutputExtremums[i], y.OutputExtremums[i]));
}
return(ret);
}
public void Normalize(double minTo, double maxTo, DataExtremumsForNetwork extremumus = null)
{
if (extremumus == null)
{
extremumus = Extremums;
}
var n = patterns.First().Input.Count;
for (int i = 0; i < n; i++)
{
Extremums.InputExtremums[i].Normalizor = new Normalizor(
extremumus.InputExtremums[i].MinValue,
extremumus.InputExtremums[i].MaxValue,
minTo,
maxTo);
}
n = patterns.First().IdealOutput.Count;
for (int i = 0; i < n; i++)
{
Extremums.OutputExtremums[i].Normalizor = new Normalizor(
extremumus.OutputExtremums[i].MinValue,
extremumus.OutputExtremums[i].MaxValue,
minTo,
maxTo);
}
foreach (Pattern p in patterns)
{
for (int i = 0; i < p.Input.Count; i++)
{
p.Input[i] = Extremums.InputExtremums[i].Normalizor.Normalize(p.Input[i]);
}
for (int i = 0; i < p.IdealOutput.Count; i++)
{
p.IdealOutput[i] = Extremums.OutputExtremums[i].Normalizor.Normalize(p.IdealOutput[i]);
}
}
}
protected int startTime; // starting time index. Needed for continuity between train and test sets
public DataSet(int startTime)
{
Extremums = new DataExtremumsForNetwork();
this.startTime = startTime;
}
