public double Run(double[][] inputs, double[] time)
{
var censor = new SurvivalOutcome[time.Length];
System.Diagnostics.Debug.Assert(censor[0] == SurvivalOutcome.Failed);
return(Run(inputs, time, censor));
}
/// <summary>
/// Computes class-label decisions for each vector in the given <paramref name="input"/>.
/// </summary>
///
/// <param name="input">The input vectors that should be classified into
/// one of the <see cref="ITransform.NumberOfOutputs"/> possible classes.</param>
///
public SurvivalOutcome[] Decide(double[][] input)
{
var result = new SurvivalOutcome[input.Length];
for (int i = 0; i < input.Length; i++)
{
result[i] = Decide(input[i]);
}
return(result);
}
/// <summary>
/// The likelihood ratio test of the overall model, also called the model chi-square test.
/// </summary>
///
/// <param name="input">A set of input data.</param>
/// <param name="time">The time-to-event before the output occurs.</param>
/// <param name="output">The corresponding output data.</param>
///
/// <remarks>
/// <para>
/// The Chi-square test, also called the likelihood ratio test or the log-likelihood test
/// is based on the deviance of the model (-2*log-likelihood). The log-likelihood ratio test
/// indicates whether there is evidence of the need to move from a simpler model to a more
/// complicated one (where the simpler model is nested within the complicated one).</para>
/// <para>
/// The difference between the log-likelihood ratios for the researcher's model and a
/// simpler model is often called the "model chi-square".</para>
/// </remarks>
///
public ChiSquareTest ChiSquare(double[][] input, double[] time, SurvivalOutcome[] output)
{
ProportionalHazards regression = new ProportionalHazards(Inputs);
double ratio = GetLogLikelihoodRatio(input, time, output, regression);
return new ChiSquareTest(ratio, Coefficients.Length);
}
/// <summary>
/// Gets the Log-Likelihood Ratio between two models.
/// </summary>
///
/// <remarks>
/// The Log-Likelihood ratio is defined as 2*(LL - LL0).
/// </remarks>
///
/// <param name="input">A set of input data.</param>
/// <param name="time">The time-to-event before the output occurs.</param>
/// <param name="output">The corresponding output data.</param>
/// <param name="hazards">Another Cox Proportional Hazards model.</param>
///
/// <returns>The Log-Likelihood ratio (a measure of performance
/// between two models) calculated over the given data sets.</returns>
///
public double GetLogLikelihoodRatio(double[][] input, double[] time, SurvivalOutcome[] output, ProportionalHazards hazards)
{
return 2.0 * (this.GetPartialLogLikelihood(input, time, output) - hazards.GetPartialLogLikelihood(input, time, output));
}
/// <summary>
/// Gets the Partial Log-Likelihood for the model.
/// </summary>
///
/// <param name="time">The time-to-event before the output occurs.</param>
/// <param name="output">The corresponding output data.</param>
///
/// <returns>
/// The Partial Log-Likelihood (a measure of performance)
/// of the model calculated over the given data set.
/// </returns>
///
public double GetPartialLogLikelihood(double[] time, SurvivalOutcome[] output)
{
double sum2 = 0;
for (int i = 0; i < time.Length; i++)
{
if (output[i] == 0)
continue;
// Compute the second sum
double sum = 0;
for (int j = 0; j < time.Length; j++)
{
if (time[j] >= time[i])
sum++;
}
sum2 += Math.Log(sum);
}
return -sum2;
}
/// <summary>
/// Gets the Partial Log-Likelihood for the model.
/// </summary>
///
/// <param name="inputs">A set of input data.</param>
/// <param name="time">The time-to-event before the output occurs.</param>
/// <param name="output">The corresponding output data.</param>
///
/// <returns>
/// The Partial Log-Likelihood (a measure of performance)
/// of the model calculated over the given data set.
/// </returns>
///
public double GetPartialLogLikelihood(double[][] inputs, double[] time, SurvivalOutcome[] output)
{
double sum1 = 0, sum2 = 0;
for (int i = 0; i < inputs.Length; i++)
{
if (output[i] == 0) continue;
// Compute the first sum
for (int j = 0; j < Coefficients.Length; j++)
sum1 += Coefficients[j] * (inputs[i][j] - Offsets[j]);
// Compute the second sum
double sum = 0;
for (int j = 0; j < inputs.Length; j++)
{
if (time[j] >= time[i])
{
double s = 0;
for (int k = 0; k < Coefficients.Length; k++)
s += Coefficients[k] * (inputs[j][k] - Offsets[k]);
sum += Math.Exp(s);
}
}
sum2 += Math.Log(sum);
}
return sum1 - sum2;
}
/// <summary>
/// Gets the Deviance for the model.
/// </summary>
///
/// <remarks>
/// The deviance is defined as -2*Log-Likelihood.
/// </remarks>
///
/// <param name="inputs">A set of input data.</param>
/// <param name="time">The time-to-event before the output occurs.</param>
/// <param name="output">The corresponding output data.</param>
///
/// <returns>
/// The deviance (a measure of performance) of the model
/// calculated over the given data sets.
/// </returns>
///
public double GetDeviance(double[][] inputs, double[] time, SurvivalOutcome[] output)
{
return -2.0 * GetPartialLogLikelihood(inputs, time, output);
}
/// <summary>
/// Initializes a new instance of the <see cref="EmpiricalHazardOptions"/> class.
/// </summary>
///
public EmpiricalHazardOptions(HazardEstimator estimator, HazardTiesMethod ties, SurvivalOutcome[] outcome)
{
Estimator = estimator;
Outcome = outcome;
Ties = ties;
}
private static void CreateExample1(out double[] times, out SurvivalOutcome[] censor)
{
// Example from http://sas-and-r.blogspot.fr/2010/05/example-738-kaplan-meier-survival.html
object[,] data =
{
// time event
{ 0.5, false },
{ 1, true },
{ 1, true },
{ 2, true },
{ 2, false },
{ 3, true },
{ 4, true },
{ 5, false },
{ 6, true },
{ 7, false },
{ 8, true },
{ 9, true },
{ 10, false },
{ 12, true },
{ 14, false },
{ 14, true },
{ 17, false },
{ 20, true },
{ 21, false },
};
times = data.GetColumn(0).To<double[]>();
censor = data.GetColumn(1).To<SurvivalOutcome[]>();
}
/// <summary>
/// Initializes a new instance of the <see cref="EmpiricalHazardOptions"/> class.
/// </summary>
///
public EmpiricalHazardOptions(HazardEstimator estimator, SurvivalOutcome[] output)
{
Estimator = estimator;
Outcome = output;
Ties = DefaultTies;
}
private void createBaseline(double[] time, SurvivalOutcome[] censor, double[] output = null)
{
if (regression.BaselineHazard == null)
return;
var hazard = regression.BaselineHazard as IFittableDistribution<double, EmpiricalHazardOptions>;
if (hazard != null)
{
// Compute an estimate of the cumulative Hazard
// function using the Nelson-Aalen estimator
hazard.Fit(time, output, new EmpiricalHazardOptions()
{
Outcome = censor,
Estimator = Estimator,
Ties = Ties
});
return;
}
var survival = regression.BaselineHazard as IFittableDistribution<double, SurvivalOptions>;
if (survival != null)
{
// Compute an estimate of the cumulative Hazard
// function using the Kaplan-Meier estimator
survival.Fit(time, new SurvivalOptions()
{
Outcome = censor,
});
}
}
/// <summary>
/// Runs the Newton-Raphson update for Cox's hazards learning until convergence.
/// </summary>
///
/// <param name="censor">The output (event) associated with each input vector.</param>
/// <param name="time">The time-to-event for the non-censored training samples.</param>
///
/// <returns>The maximum relative change in the parameters after the iteration.</returns>
///
public double Run(double[] time, SurvivalOutcome[] censor)
{
if (time.Length != censor.Length)
{
throw new DimensionMismatchException("time",
"The time and output vector must have the same length.");
}
// Sort data by time to accelerate performance
EmpiricalHazardDistribution.Sort(ref time, ref censor);
createBaseline(time, censor);
return regression.GetPartialLogLikelihood(time, censor);
}
/// <summary>
/// Runs the Newton-Raphson update for Cox's hazards learning until convergence.
/// </summary>
///
/// <param name="inputs">The input data.</param>
/// <param name="censor">The output (event) associated with each input vector.</param>
/// <param name="time">The time-to-event for the non-censored training samples.</param>
///
/// <returns>The maximum relative change in the parameters after the iteration.</returns>
///
public double Run(double[][] inputs, double[] time, SurvivalOutcome[] censor)
{
if (inputs.Length != time.Length || time.Length != censor.Length)
{
throw new DimensionMismatchException("time",
"The inputs, time and output vector must have the same length.");
}
// Sort data by time to accelerate performance
EmpiricalHazardDistribution.Sort(ref time, ref censor, ref inputs);
double[] means = new double[parameterCount];
double[] sdev = new double[parameterCount];
for (int i = 0; i < sdev.Length; i++)
sdev[i] = 1;
if (normalize)
{
// Store means as regression centers
means = inputs.Mean();
for (int i = 0; i < means.Length; i++)
regression.Offsets[i] = means[i];
// Convert to unit scores for increased accuracy
sdev = Accord.Statistics.Tools.StandardDeviation(inputs);
inputs = inputs.Subtract(means, 0).ElementwiseDivide(sdev, 0, inPlace: true);
for (int i = 0; i < regression.Coefficients.Length; i++)
regression.Coefficients[i] *= sdev[i];
}
// Compute actual outputs
double[] output = new double[inputs.Length];
for (int i = 0; i < output.Length; i++)
{
double sum = 0;
for (int j = 0; j < regression.Coefficients.Length; j++)
sum += regression.Coefficients[j] * inputs[i][j];
output[i] = Math.Exp(sum);
}
// Compute ties
int[] ties = new int[inputs.Length];
for (int i = 0; i < inputs.Length; i++)
for (int j = 0; j < time.Length; j++)
if (time[j] == time[i]) ties[i]++;
if (parameterCount == 0)
{
createBaseline(time, censor, output);
return regression.GetPartialLogLikelihood(inputs, time, censor);
}
CurrentIteration = 0;
double smooth = Lambda;
do
{
// learning iterations until convergence
// or maximum number of iterations reached
CurrentIteration++;
// Reset Hessian matrix and gradient
Array.Clear(gradient, 0, gradient.Length);
Array.Clear(hessian, 0, hessian.Length);
// For each observation instance
for (int i = 0; i < inputs.Length; i++)
{
// Check if we should censor
if (censor[i] == SurvivalOutcome.Censored)
continue;
// Compute partials
double den = 0;
Array.Clear(partialGradient, 0, partialGradient.Length);
Array.Clear(partialHessian, 0, partialHessian.Length);
for (int j = 0; j < inputs.Length; j++)
{
if (time[j] >= time[i])
den += output[j];
}
for (int j = 0; j < inputs.Length; j++)
{
if (time[j] >= time[i])
{
// Compute partial gradient
for (int k = 0; k < partialGradient.Length; k++)
partialGradient[k] += inputs[j][k] * output[j] / den;
// Compute partial Hessian
for (int ii = 0; ii < inputs[j].Length; ii++)
for (int jj = 0; jj < inputs[j].Length; jj++)
partialHessian[ii, jj] += inputs[j][ii] * inputs[j][jj] * output[j] / den;
}
}
// Compute gradient vector
for (int j = 0; j < gradient.Length; j++)
gradient[j] += inputs[i][j] - partialGradient[j];
// Compute Hessian matrix
for (int j = 0; j < partialGradient.Length; j++)
for (int k = 0; k < partialGradient.Length; k++)
hessian[j, k] -= partialHessian[j, k] - partialGradient[j] * partialGradient[k];
}
// Decompose to solve the linear system. Usually the Hessian will
// be invertible and LU will succeed. However, sometimes the Hessian
// may be singular and a Singular Value Decomposition may be needed.
// The SVD is very stable, but is quite expensive, being on average
// about 10-15 times more expensive than LU decomposition. There are
// other ways to avoid a singular Hessian. For a very interesting
// reading on the subject, please see:
//
// - Jeff Gill & Gary King, "What to Do When Your Hessian Is Not Invertible",
// Sociological Methods & Research, Vol 33, No. 1, August 2004, 54-87.
// Available in: http://gking.harvard.edu/files/help.pdf
//
decomposition = new SingularValueDecomposition(hessian);
double[] deltas = decomposition.Solve(gradient);
if (convergence.Iterations > 0 || convergence.Tolerance > 0)
{
// Update coefficients using the calculated deltas
for (int i = 0; i < regression.Coefficients.Length; i++)
regression.Coefficients[i] -= smooth * deltas[i];
}
smooth += Lambda;
if (smooth > 1)
smooth = 1;
// Check relative maximum parameter change
convergence.NewValues = regression.Coefficients;
if (convergence.HasDiverged)
{
// Restore previous coefficients
for (int i = 0; i < regression.Coefficients.Length; i++)
regression.Coefficients[i] = convergence.OldValues[i];
}
// Recompute current outputs
for (int i = 0; i < output.Length; i++)
{
double sum = 0;
for (int j = 0; j < regression.Coefficients.Length; j++)
sum += regression.Coefficients[j] * inputs[i][j];
output[i] = Math.Exp(sum);
}
} while (!convergence.HasConverged);
for (int i = 0; i < regression.Coefficients.Length; i++)
regression.Coefficients[i] /= sdev[i];
if (computeStandardErrors)
{
// Grab the regression information matrix
double[,] inverse = decomposition.Inverse();
// Calculate coefficients' standard errors
double[] standardErrors = regression.StandardErrors;
for (int i = 0; i < standardErrors.Length; i++)
standardErrors[i] = Math.Sqrt(Math.Abs(inverse[i, i])) / sdev[i];
}
if (computeBaselineFunction)
createBaseline(time, censor, output);
return regression.GetPartialLogLikelihood(inputs, time, censor);
}
public double Run(double[][] inputs, double[] time)
{
var censor = new SurvivalOutcome[time.Length];
System.Diagnostics.Debug.Assert(censor[0] == SurvivalOutcome.Failed);
return Run(inputs, time, censor);
}
