using System;

using System.Collections.Generic;

using System.Linq;

using System.Text;

namespace NumSharp.Extensions

{

public static partial class NDArrayExtensions

{

public static NDArray_Legacy<double> Std(this NDArray_Legacy<NDArray_Legacy<double>> np, int axis = -1)

{

var std = new NDArray_Legacy<double>();

var mean = np.Mean(axis);

// axis == -1: DEFAULT; to compute the standard deviation of the flattened array.

if (axis == -1)

{

var sum = np.Data.Select(d => d.Data.Select(p => Math.Pow(Math.Abs(p - mean.Data[0]), 2)).Sum()).Sum();

std.Data.Add(Math.Sqrt(sum / np.Size));

}

// to compute mean by compressing row and row

else if (axis == 0)

{

double[] sumVec = new double[np.Data[0].Length];

for (int d = 0; d < np.Length; d++)

{

for (int p = 0; p < np.Data[0].Length; p++)

{

sumVec[p] += np.Data[d][p];

}

}

for (int d = 0; d < np.Data[0].Length; d++)

{

mean.Data.Add(sumVec[d] / np.Length);

}

}

else if (axis == 1)

{

for (int d = 0; d < np.Length; d++)

{

double rowSum = 0;

for (int p = 0; p < np.Data[0].Length; p++)

{

rowSum += np.Data[d][p];

}

mean.Data.Add(rowSum / np.Data[0].Length);

}

}

return std;

}

}

}