autoTableOfReal FormantTier_downto_TableOfReal (FormantTier me, int includeFormants, int includeBandwidths) {
try {
int maximumNumberOfFormants = FormantTier_getMaxNumFormants (me);
autoTableOfReal thee = TableOfReal_create (my points -> size, 1 +
( includeFormants ? maximumNumberOfFormants : 0 ) +
( includeBandwidths ? maximumNumberOfFormants : 0 ));
TableOfReal_setColumnLabel (thee.peek(), 1, U"Time");
for (long icol = 1, iformant = 1; iformant <= maximumNumberOfFormants; iformant ++) {
char32 label [4];
if (includeFormants) {
Melder_sprint (label,4, U"F", iformant);
TableOfReal_setColumnLabel (thee.peek(), ++ icol, label);
}
if (includeBandwidths) {
Melder_sprint (label,4, U"B", iformant);
TableOfReal_setColumnLabel (thee.peek(), ++ icol, label);
}
}
for (long ipoint = 1; ipoint <= my points -> size; ipoint ++) {
FormantPoint point = (FormantPoint) my points -> item [ipoint];
thy data [ipoint] [1] = point -> time;
for (long icol = 1, iformant = 1; iformant <= maximumNumberOfFormants; iformant ++) {
if (includeFormants) thy data [ipoint] [++ icol] = point -> formant [iformant-1];
if (includeBandwidths) thy data [ipoint] [++ icol] = point -> bandwidth [iformant-1];
}
}
return thee;
} catch (MelderError) {
Melder_throw (me, U": not converted to TableOfReal.");
}
}
autoConfusion Confusion_groupResponses (Confusion me, const char32 *labels, const char32 *newLabel, long newpos) {
try {
long ncondense = Melder_countTokens (labels);
autoNUMvector<long> icol (1, my numberOfColumns);
for (long i = 1; i <= my numberOfColumns; i++) {
icol[i] = i;
}
for (char32 *token = Melder_firstToken (labels); token != 0; token = Melder_nextToken ()) {
for (long i = 1; i <= my numberOfColumns; i++) {
if (Melder_equ (token, my columnLabels[i])) {
icol[i] = 0;
break;
}
}
}
long nfound = 0;
for (long i = 1; i <= my numberOfColumns; i++) {
if (icol[i] == 0) {
nfound ++;
}
}
if (nfound == 0) {
Melder_throw (U"Invalid response labels.");
}
if (nfound != ncondense) {
Melder_warning (U"One or more of the given response labels are suspect.");
}
long newnresp = my numberOfColumns - nfound + 1;
if (newpos < 1) {
newpos = 1;
}
if (newpos > newnresp) {
newpos = newnresp;
}
autoConfusion thee = Confusion_create (my numberOfRows, newnresp);
NUMstrings_copyElements (my rowLabels, thy rowLabels, 1, my numberOfRows);
TableOfReal_setColumnLabel (thee.get(), newpos, newLabel);
long inewcol = 1;
for (long i = 1; i <= my numberOfColumns; i++) {
long colpos = newpos;
if (icol[i] > 0) {
if (inewcol == newpos) {
inewcol++;
}
colpos = inewcol;
inewcol++;
TableOfReal_setColumnLabel (thee.get(), colpos, my columnLabels[i]);
}
for (long j = 1; j <= my numberOfRows; j++) {
thy data[j][colpos] += my data[j][i];
}
}
return thee;
} catch (MelderError) {
Melder_throw (me, U": responses not grouped.");
}
}
TableOfReal RealTier_downto_TableOfReal (I, const wchar_t *timeLabel, const wchar_t *valueLabel) {
iam (RealTier);
TableOfReal thee = TableOfReal_create (my points -> size, 2); cherror
TableOfReal_setColumnLabel (thee, 1, timeLabel); cherror
TableOfReal_setColumnLabel (thee, 2, valueLabel); cherror
for (long i = 1; i <= my points -> size; i ++) {
RealPoint point = (structRealPoint *)my points -> item [i];
thy data [i] [1] = point -> time;
thy data [i] [2] = point -> value;
}
end:
iferror forget (thee);
return thee;
}
autoTableOfReal RealTier_downto_TableOfReal (RealTier me, const char32 *timeLabel, const char32 *valueLabel) {
try {
autoTableOfReal thee = TableOfReal_create (my points.size, 2);
TableOfReal_setColumnLabel (thee.get(), 1, timeLabel);
TableOfReal_setColumnLabel (thee.get(), 2, valueLabel);
for (long i = 1; i <= my points.size; i ++) {
RealPoint point = my points.at [i];
thy data [i] [1] = point -> number;
thy data [i] [2] = point -> value;
}
return thee;
} catch (MelderError) {
Melder_throw (me, U": not converted to TableOfReal.");
}
}
CrossCorrelationTable EEG_to_CrossCorrelationTable (EEG me, double startTime, double endTime, double lagTime, const wchar_t *channelRanges)
{
try {
// autowindow
if (startTime == endTime) {
startTime = my xmin; endTime = my xmax;
}
// don't allow times outside domain
if (startTime < my xmin) {
startTime = my xmin;
}
if (endTime > my xmax) {
endTime = my xmax;
}
autoEEG thee = my f_extractPart (startTime, endTime, true);
long numberOfChannels;
autoNUMvector <long> channels (NUMstring_getElementsOfRanges (channelRanges, thy d_numberOfChannels, & numberOfChannels, NULL, L"channel", true), 1);
autoSound soundPart = Sound_copyChannelRanges (thy d_sound, channelRanges);
autoCrossCorrelationTable him = Sound_to_CrossCorrelationTable (soundPart.peek(), startTime, endTime, lagTime);
// assign channel names
for (long i = 1; i <= numberOfChannels; i++) {
long ichannel = channels[i];
wchar_t *label = my d_channelNames[ichannel];
TableOfReal_setRowLabel (him.peek(), i, label);
TableOfReal_setColumnLabel (him.peek(), i, label);
}
return him.transfer();
} catch (MelderError) {
Melder_throw (me, ": no CrossCorrelationTable calculated.");
}
}
autoTableOfReal FFNet_extractWeights (FFNet me, long layer) {
try {
FFNet_checkLayerNumber (me, layer);
long numberOfUnitsFrom = my nUnitsInLayer[layer - 1] + 1;
long numberOfUnitsTo = my nUnitsInLayer[layer];
autoTableOfReal thee = TableOfReal_create (numberOfUnitsFrom, numberOfUnitsTo);
char32 label[40];
for (long i = 1; i <= numberOfUnitsFrom - 1; i++) {
Melder_sprint (label,40, U"L", layer - 1, U"-", i);
TableOfReal_setRowLabel (thee.peek(), i, label);
}
TableOfReal_setRowLabel (thee.peek(), numberOfUnitsFrom, U"Bias");
for (long i = 1; i <= numberOfUnitsTo; i++) {
Melder_sprint (label,40, U"L", layer, U"-", i);
TableOfReal_setColumnLabel (thee.peek(), i, label);
}
long node = 1;
for (long i = 0; i < layer; i++) {
node += my nUnitsInLayer[i] + 1;
}
for (long i = 1; i <= numberOfUnitsTo; i++, node++) {
long k = 1;
for (long j = my wFirst[node]; j <= my wLast[node]; j++) {
thy data[k++][i] = my w[j];
}
}
return thee;
} catch (MelderError) {
Melder_throw (me, U": no TableOfReal created.");
}
}
Correlation ClassificationTable_to_Correlation_columns (ClassificationTable me) {
try {
autoCorrelation thee = Correlation_create (my numberOfColumns);
for (long icol = 1; icol <= thy numberOfColumns; icol++) {
char32 *label = my columnLabels[icol];
TableOfReal_setRowLabel (thee.peek(), icol, label);
TableOfReal_setColumnLabel (thee.peek(), icol, label);
}
for (long irow = 1; irow <= thy numberOfColumns; irow++) {
thy data[irow][irow] = 1.0;
for (long icol = irow + 1; icol <= thy numberOfColumns; icol++) {
double n11 = 0.0, n22 = 0.0, n12 = 0.0;
for (long i = 1; i <= my numberOfRows; i++) {
n12 += my data[i][irow] * my data[i][icol];
n11 += my data[i][irow] * my data[i][irow];
n22 += my data[i][icol] * my data[i][icol];
}
// probabilities might be very low!
if (n12 > 0.0 && n22 > 0.0) {
thy data[irow][icol] = thy data[icol][irow] = n12 / sqrt (n11 * n22);
}
}
}
thy numberOfObservations = my numberOfRows;
return thee.transfer();
} catch (MelderError) {
Melder_throw (me, U": no correlation created.");
}
}
autoConfusion Categories_to_Confusion (Categories me, Categories thee) {
try {
if (my size != thy size) {
Melder_throw (U"Categories_to_Confusion: dimensions do not agree.");
}
autoCategories ul1 = Categories_selectUniqueItems (me);
autoCategories ul2 = Categories_selectUniqueItems (thee);
autoConfusion him = Confusion_create (ul1->size, ul2->size);
for (long i = 1; i <= ul1->size; i ++) {
SimpleString s = ul1->at [i];
TableOfReal_setRowLabel (him.get(), i, s -> string);
}
for (long i = 1; i <= ul2->size; i ++) {
SimpleString s = ul2->at [i];
TableOfReal_setColumnLabel (him.get(), i, s -> string);
}
for (long i = 1; i <= my size; i ++) {
SimpleString myi = my at [i], thyi = thy at [i];
Confusion_increase (him.get(), SimpleString_c (myi), SimpleString_c (thyi));
}
return him;
} catch (MelderError) {
Melder_throw (me, U": no Confusion created.");
}
}
/*
raw r[j]: eigenvec[i][j]
unstandardized u[j]: sqrt(N-g) * r[j]
standardized s[j]: u[j] sqrt (w[i][i] / (N-g))
*/
autoTableOfReal Discriminant_extractCoefficients (Discriminant me, int choice) {
try {
int raw = choice == 0, standardized = choice == 2;
long nx = my eigen -> dimension, ny = my eigen -> numberOfEigenvalues;
SSCP total = my total.get();
autoTableOfReal thee = TableOfReal_create (ny, nx + 1);
NUMstrings_copyElements (my total -> columnLabels, thy columnLabels, 1, nx);
autoSSCP within;
if (standardized) {
within = Discriminant_extractPooledWithinGroupsSSCP (me);
}
TableOfReal_setColumnLabel (thee.get(), nx + 1, U"constant");
TableOfReal_setSequentialRowLabels (thee.get(), 1, ny, U"function_", 1, 1);
double scale = sqrt (total -> numberOfObservations - my numberOfGroups);
double *centroid = my total -> centroid;
for (long i = 1; i <= ny; i++) {
double u0 = 0.0, ui;
for (long j = 1; j <= nx; j++) {
if (standardized) {
scale = sqrt (within -> data[j][j]);
}
thy data[i][j] = ui = scale * my eigen -> eigenvectors[i][j];;
u0 += ui * centroid[j];
}
thy data[i][nx + 1] = raw ? 0.0 : -u0;
}
return thee;
} catch (MelderError) {
Melder_throw (me, U": coefficients not extracted.");
}
}
Any TablesOfReal_appendMany (Collection me) {
try {
if (my size == 0) Melder_throw ("Cannot add zero tables.");
TableOfReal thee = static_cast <TableOfReal> (my item [1]);
long totalNumberOfRows = thy numberOfRows;
long numberOfColumns = thy numberOfColumns;
for (long itab = 2; itab <= my size; itab ++) {
thee = static_cast <TableOfReal> (my item [itab]);
totalNumberOfRows += thy numberOfRows;
if (thy numberOfColumns != numberOfColumns) Melder_throw ("Numbers of columns do not match.");
}
autoTableOfReal him = static_cast <TableOfReal> (_Thing_new (thy classInfo));
TableOfReal_init (him.peek(), totalNumberOfRows, numberOfColumns);
/* Unsafe: new attributes not initialized. */
for (long icol = 1; icol <= numberOfColumns; icol ++) {
TableOfReal_setColumnLabel (him.peek(), icol, thy columnLabels [icol]);
}
totalNumberOfRows = 0;
for (long itab = 1; itab <= my size; itab ++) {
thee = static_cast <TableOfReal> (my item [itab]);
for (long irow = 1; irow <= thy numberOfRows; irow ++) {
totalNumberOfRows ++;
TableOfReal_setRowLabel (him.peek(), totalNumberOfRows, thy rowLabels [irow]);
for (long icol = 1; icol <= numberOfColumns; icol ++)
his data [totalNumberOfRows] [icol] = thy data [irow] [icol];
}
}
Melder_assert (totalNumberOfRows == his numberOfRows);
return him.transfer();
} catch (MelderError) {
Melder_throw ("TableOfReal objects not appended.");
}
}
Any TablesOfReal_append (TableOfReal me, TableOfReal thee) {
try {
if (thy numberOfColumns != my numberOfColumns)
Melder_throw (L"Numbers of columns are ", my numberOfColumns, " and ", thy numberOfColumns, " but should be equal.");
autoTableOfReal him = static_cast <TableOfReal> (_Thing_new (my classInfo));
TableOfReal_init (him.peek(), my numberOfRows + thy numberOfRows, my numberOfColumns);
/* Unsafe: new attributes not initialized. */
for (long icol = 1; icol <= my numberOfColumns; icol ++) {
TableOfReal_setColumnLabel (him.peek(), icol, my columnLabels [icol]);
}
for (long irow = 1; irow <= my numberOfRows; irow ++) {
TableOfReal_setRowLabel (him.peek(), irow, my rowLabels [irow]);
for (long icol = 1; icol <= my numberOfColumns; icol ++)
his data [irow] [icol] = my data [irow] [icol];
}
for (long irow = 1; irow <= thy numberOfRows; irow ++) {
long hisRow = irow + my numberOfRows;
TableOfReal_setRowLabel (him.peek(), hisRow, thy rowLabels [irow]);
for (long icol = 1; icol <= my numberOfColumns; icol ++)
his data [hisRow] [icol] = thy data [irow] [icol];
}
return him.transfer();
} catch (MelderError) {
Melder_throw ("TableOfReal objects not appended.");
}
}
TableOfReal Table_to_TableOfReal (Table me, long labelColumn) {
try {
if (labelColumn < 1 || labelColumn > my numberOfColumns) labelColumn = 0;
autoTableOfReal thee = TableOfReal_create (my rows -> size, labelColumn ? my numberOfColumns - 1 : my numberOfColumns);
for (long icol = 1; icol <= my numberOfColumns; icol ++) {
Table_numericize_Assert (me, icol);
}
if (labelColumn) {
for (long icol = 1; icol < labelColumn; icol ++) {
TableOfReal_setColumnLabel (thee.peek(), icol, my columnHeaders [icol]. label);
}
for (long icol = labelColumn + 1; icol <= my numberOfColumns; icol ++) {
TableOfReal_setColumnLabel (thee.peek(), icol - 1, my columnHeaders [icol]. label);
}
for (long irow = 1; irow <= my rows -> size; irow ++) {
TableRow row = static_cast <TableRow> (my rows -> item [irow]);
wchar_t *string = row -> cells [labelColumn]. string;
TableOfReal_setRowLabel (thee.peek(), irow, string ? string : L"");
for (long icol = 1; icol < labelColumn; icol ++) {
thy data [irow] [icol] = row -> cells [icol]. number; // Optimization.
//thy data [irow] [icol] = Table_getNumericValue_Assert (me, irow, icol);
}
for (long icol = labelColumn + 1; icol <= my numberOfColumns; icol ++) {
thy data [irow] [icol - 1] = row -> cells [icol]. number; // Optimization.
//thy data [irow] [icol - 1] = Table_getNumericValue_Assert (me, irow, icol);
}
}
} else {
for (long icol = 1; icol <= my numberOfColumns; icol ++) {
TableOfReal_setColumnLabel (thee.peek(), icol, my columnHeaders [icol]. label);
}
for (long irow = 1; irow <= my rows -> size; irow ++) {
TableRow row = static_cast <TableRow> (my rows -> item [irow]);
for (long icol = 1; icol <= my numberOfColumns; icol ++) {
thy data [irow] [icol] = row -> cells [icol]. number; // Optimization.
//thy data [irow] [icol] = Table_getNumericValue_Assert (me, irow, icol);
}
}
}
return thee.transfer();
} catch (MelderError) {
Melder_throw (me, ": not converted to TableOfReal.");
}
}
TableOfReal AffineTransform_extractMatrix (I) {
iam (AffineTransform);
try {
autoTableOfReal thee = TableOfReal_create (my n, my n);
NUMmatrix_copyElements (my r, thy data, 1, my n, 1, my n);
for (long i = 1; i <= my n; i++) {
wchar_t label[20];
(void) swprintf (label, 20, L"%ld", i);
TableOfReal_setRowLabel (thee.peek(), i, label);
TableOfReal_setColumnLabel (thee.peek(), i, label);
}
return thee.transfer();
} catch (MelderError) {
Melder_throw (me, ": transformation matrix not extracted.");
}
}
autoConfusion Confusion_createFromStringses (Strings me, Strings thee) {
try {
if (my numberOfStrings < 1 || thy numberOfStrings < 1) {
Melder_throw (U"Empty Strings.");
}
autoConfusion him = Confusion_create (my numberOfStrings, thy numberOfStrings);
for (long irow = 1; irow <= my numberOfStrings; irow++) {
const char32 *label = my strings[irow];
TableOfReal_setRowLabel (him.get(), irow, label);
}
for (long icol = 1; icol <= thy numberOfStrings; icol++) {
const char32 *label = thy strings[icol];
TableOfReal_setColumnLabel (him.get(), icol, label);
}
return him;
} catch (MelderError) {
Melder_throw (me, U": could not create Confusion with ", thee);
}
}
EditDistanceTable EditDistanceTable_create (Strings target, Strings source) {
try {
autoEditDistanceTable me = Thing_new (EditDistanceTable);
long numberOfSourceSymbols = source -> numberOfStrings, numberOfTargetSymbols = target -> numberOfStrings;
TableOfReal_init (me.peek(), numberOfTargetSymbols + 1, numberOfSourceSymbols + 1);
TableOfReal_setColumnLabel (me.peek(), 1, U"");
for (long j = 1; j <= numberOfSourceSymbols; j++) {
my columnLabels[j + 1] = Melder_dup (source -> strings[j]);
}
TableOfReal_setRowLabel (me.peek(), 1, U"");
for (long i = 1; i <= numberOfTargetSymbols; i++) {
my rowLabels[i + 1] = Melder_dup (target -> strings[i]);
}
my warpingPath = WarpingPath_create (numberOfTargetSymbols + numberOfSourceSymbols + 1);
my editCostsTable = EditCostsTable_createDefault ();
EditDistanceTable_findPath (me.peek(), 0);
return me.transfer();
} catch (MelderError) {
Melder_throw (U"EditDistanceTable not created.");
}
}
autoConfusion Confusion_createSimple (const char32 *labels) {
try {
long numberOfLabels = Melder_countTokens (labels);
if (numberOfLabels < 1) {
Melder_throw (U"Not enough labels.");
}
autoConfusion me = Confusion_create (numberOfLabels, numberOfLabels);
long ilabel = 1;
for (char32 *token = Melder_firstToken (labels); token != 0; token = Melder_nextToken ()) {
for (long i = 1; i <= ilabel - 1; i++) {
if (Melder_cmp (token, my rowLabels[i]) == 0) {
Melder_throw (U"Label ", i, U"and ", ilabel, U"may not be equal.");
}
}
TableOfReal_setRowLabel (me.get(), ilabel, token);
TableOfReal_setColumnLabel (me.get(), ilabel, token);
ilabel++;
}
return me;
} catch (MelderError) {
Melder_throw (U"Simple Confusion not created.");
}
}
TableOfReal FFNet_extractWeights (FFNet me, long layer)
{
TableOfReal thee = NULL;
long i, node = 1, numberOfUnitsFrom, numberOfUnitsTo;
wchar_t label[20];
if (! FFNet_checkLayerNumber (me, layer)) return NULL;
numberOfUnitsFrom = my nUnitsInLayer[layer-1] + 1;
numberOfUnitsTo = my nUnitsInLayer[layer];
thee = TableOfReal_create (numberOfUnitsFrom, numberOfUnitsTo);
if (thee == NULL) return NULL;
for (i = 1; i <= numberOfUnitsFrom - 1; i++)
{
swprintf (label,20,L"L%ld-%ld", layer-1, i);
TableOfReal_setRowLabel (thee, i, label);
}
TableOfReal_setRowLabel (thee, numberOfUnitsFrom, L"Bias");
for (i = 1; i <= numberOfUnitsTo; i++)
{
swprintf (label,20,L"L%ld-%ld", layer, i);
TableOfReal_setColumnLabel (thee, i, label);
}
for (i = 0; i < layer; i++) node += my nUnitsInLayer[i] + 1;
for (i = 1; i <= numberOfUnitsTo; i++, node++)
{
long j, k = 1;
for (j = my wFirst[node]; j <= my wLast[node]; j++)
{
thy data[k++][i] = my w[j];
}
}
return thee;
}
ClassificationTable Discriminant_and_TableOfReal_to_ClassificationTable (Discriminant me, TableOfReal thee,
int poolCovarianceMatrices, int useAprioriProbabilities) {
try {
long g = Discriminant_getNumberOfGroups (me);
long p = Eigen_getDimensionOfComponents (me);
long m = thy numberOfRows;
if (p != thy numberOfColumns) Melder_throw
(U"The number of columns does not agree with the dimension of the discriminant.");
autoNUMvector<double> log_p (1, g);
autoNUMvector<double> log_apriori (1, g);
autoNUMvector<double> ln_determinant (1, g);
autoNUMvector<double> buf (1, p);
autoNUMvector<SSCP> sscpvec (1, g);
autoSSCP pool = SSCPs_to_SSCP_pool (my groups);
autoClassificationTable him = ClassificationTable_create (m, g);
NUMstrings_copyElements (thy rowLabels, his rowLabels, 1, m);
// Scale the sscp to become a covariance matrix.
for (long i = 1; i <= p; i++) {
for (long k = i; k <= p; k++) {
pool -> data[k][i] = (pool -> data[i][k] /= (pool -> numberOfObservations - g));
}
}
double lnd;
autoSSCPs agroups = 0; SSCPs groups;
if (poolCovarianceMatrices) {
/*
Covariance matrix S can be decomposed as S = L.L'. Calculate L^-1.
L^-1 will be used later in the Mahalanobis distance calculation:
v'.S^-1.v == v'.L^-1'.L^-1.v == (L^-1.v)'.(L^-1.v).
*/
NUMlowerCholeskyInverse (pool -> data, p, &lnd);
for (long j = 1; j <= g; j++) {
ln_determinant[j] = lnd;
sscpvec[j] = pool.peek();
}
groups = (SSCPs) my groups;
} else {
// Calculate the inverses of all group covariance matrices.
// In case of a singular matrix, substitute inverse of pooled.
agroups.reset (Data_copy ( (SSCPs) my groups)); groups = agroups.peek();
long npool = 0;
for (long j = 1; j <= g; j++) {
SSCP t = (SSCP) groups -> item[j];
long no = (long) floor (SSCP_getNumberOfObservations (t));
for (long i = 1; i <= p; i++) {
for (long k = i; k <= p; k++) {
t -> data[k][i] = (t -> data[i][k] /= (no - 1));
}
}
sscpvec[j] = (SSCP) groups -> item[j];
try {
NUMlowerCholeskyInverse (t -> data, p, &ln_determinant[j]);
} catch (MelderError) {
// Try the alternative: the pooled covariance matrix.
// Clear the error.
Melder_clearError ();
if (npool == 0) {
NUMlowerCholeskyInverse (pool -> data, p, &lnd);
}
npool++;
sscpvec[j] = pool.peek();
ln_determinant[j] = lnd;
}
}
if (npool > 0) {
Melder_warning (npool, U" groups use pooled covariance matrix.");
}
}
// Labels for columns in ClassificationTable
for (long j = 1; j <= g; j++) {
const char32 *name = Thing_getName ( (Thing) my groups -> item[j]);
if (! name) {
name = U"?";
}
TableOfReal_setColumnLabel (him.peek(), j, name);
}
// Normalize the sum of the apriori probabilities to 1.
// Next take ln (p) because otherwise probabilities might be too small to represent.
NUMvector_normalize1 (my aprioriProbabilities, g);
double logg = log (g);
for (long j = 1; j <= g; j++) {
log_apriori[j] = useAprioriProbabilities ? log (my aprioriProbabilities[j]) : - logg;
}
// Generalized squared distance function:
// D^2(x) = (x - mu)' S^-1 (x - mu) + ln (determinant(S)) - 2 ln (apriori)
for (long i = 1; i <= m; i++) {
double norm = 0, pt_max = -1e38;
for (long j = 1; j <= g; j++) {
SSCP t = (SSCP) groups -> item[j];
double md = mahalanobisDistanceSq (sscpvec[j] -> data, p, thy data[i], t -> centroid, buf.peek());
double pt = log_apriori[j] - 0.5 * (ln_determinant[j] + md);
if (pt > pt_max) {
pt_max = pt;
}
log_p[j] = pt;
}
for (long j = 1; j <= g; j++) {
norm += (log_p[j] = exp (log_p[j] - pt_max));
}
for (long j = 1; j <= g; j++) {
his data[i][j] = log_p[j] / norm;
}
}
return him.transfer();
} catch (MelderError) {
Melder_throw (U"ClassificationTable from Discriminant & TableOfReal not created.");
}
}
autoClassificationTable Discriminant_and_TableOfReal_to_ClassificationTable_dw (Discriminant me, TableOfReal thee, int poolCovarianceMatrices, int useAprioriProbabilities, double alpha, double minProb, autoTableOfReal *displacements) {
try {
long g = Discriminant_getNumberOfGroups (me);
long p = Eigen_getDimensionOfComponents (my eigen.get());
long m = thy numberOfRows;
if (p != thy numberOfColumns) Melder_throw
(U"The number of columns does not agree with the dimension of the discriminant.");
autoNUMvector<double> log_p (1, g);
autoNUMvector<double> log_apriori (1, g);
autoNUMvector<double> ln_determinant (1, g);
autoNUMvector<double> buf (1, p);
autoNUMvector<double> displacement (1, p);
autoNUMvector<double> x (1, p);
autoNUMvector<SSCP> sscpvec (1, g);
autoSSCP pool = SSCPList_to_SSCP_pool (my groups.get());
autoClassificationTable him = ClassificationTable_create (m, g);
NUMstrings_copyElements (thy rowLabels, his rowLabels, 1, m);
autoTableOfReal adisplacements = Data_copy (thee);
// Scale the sscp to become a covariance matrix.
for (long i = 1; i <= p; i ++) {
for (long k = i; k <= p; k ++) {
pool -> data [k] [i] = pool -> data [i] [k] /= pool -> numberOfObservations - g;
}
}
double lnd;
autoSSCPList agroups;
SSCPList groups;
if (poolCovarianceMatrices) {
// Covariance matrix S can be decomposed as S = L.L'. Calculate L^-1.
// L^-1 will be used later in the Mahalanobis distance calculation:
// v'.S^-1.v == v'.L^-1'.L^-1.v == (L^-1.v)'.(L^-1.v).
NUMlowerCholeskyInverse (pool -> data, p, & lnd);
for (long j = 1; j <= g; j ++) {
ln_determinant [j] = lnd;
sscpvec [j] = pool.get();
}
groups = my groups.get();
} else {
//Calculate the inverses of all group covariance matrices.
// In case of a singular matrix, substitute inverse of pooled.
agroups = Data_copy (my groups.get());
groups = agroups.get();
long npool = 0;
for (long j = 1; j <= g; j ++) {
SSCP t = groups->at [j];
long no = (long) floor (SSCP_getNumberOfObservations (t));
for (long i = 1; i <= p; i ++) {
for (long k = i; k <= p; k ++) {
t -> data [k] [i] = t -> data [i] [k] /= no - 1;
}
}
sscpvec [j] = groups->at [j];
try {
NUMlowerCholeskyInverse (t -> data, p, & ln_determinant [j]);
} catch (MelderError) {
// Try the alternative: the pooled covariance matrix.
// Clear the error.
Melder_clearError ();
if (npool == 0) {
NUMlowerCholeskyInverse (pool -> data, p, & lnd);
}
npool ++;
sscpvec [j] = pool.get();
ln_determinant [j] = lnd;
}
}
if (npool > 0) {
Melder_warning (npool, U" groups use pooled covariance matrix.");
}
}
// Labels for columns in ClassificationTable
for (long j = 1; j <= g; j ++) {
const char32 *name = Thing_getName (my groups->at [j]);
if (! name) {
name = U"?";
}
TableOfReal_setColumnLabel (him.get(), j, name);
}
// Normalize the sum of the apriori probabilities to 1.
// Next take ln (p) because otherwise probabilities might be too small to represent.
double logg = log (g);
NUMvector_normalize1 (my aprioriProbabilities, g);
for (long j = 1; j <= g; j ++) {
log_apriori[j] = ( useAprioriProbabilities ? log (my aprioriProbabilities[j]) : - logg );
}
// Generalized squared distance function:
// D^2(x) = (x - mu)' S^-1 (x - mu) + ln (determinant(S)) - 2 ln (apriori)
for (long i = 1; i <= m; i ++) {
SSCP winner;
double norm = 0, pt_max = -1e308;
long iwinner = 1;
for (long k = 1; k <= p; k ++) {
x [k] = thy data [i] [k] + displacement [k];
}
for (long j = 1; j <= g; j ++) {
SSCP t = groups->at [j];
double md = mahalanobisDistanceSq (sscpvec [j] -> data, p, x.peek(), t -> centroid, buf.peek());
double pt = log_apriori [j] - 0.5 * (ln_determinant [j] + md);
if (pt > pt_max) {
pt_max = pt;
iwinner = j;
}
log_p [j] = pt;
}
for (long j = 1; j <= g; j ++) {
norm += log_p [j] = exp (log_p [j] - pt_max);
}
for (long j = 1; j <= g; j ++) {
his data [i] [j] = log_p [j] / norm;
}
// Save old displacement, calculate new displacement
winner = groups->at [iwinner];
for (long k = 1; k <= p; k ++) {
adisplacements -> data [i] [k] = displacement [k];
if (his data [i] [iwinner] > minProb) {
double delta_k = winner -> centroid [k] - x [k];
displacement [k] += alpha * delta_k;
}
}
}
*displacements = adisplacements.move();
return him;
} catch (MelderError) {
Melder_throw (U"ClassificationTable for Weenink procedure not created.");
}
}
TableOfReal TableOfReal_readFromHeaderlessSpreadsheetFile (MelderFile file) {
try {
autostring string = MelderFile_readText (file);
long nrow, ncol, nelements;
/*
* Count columns.
*/
ncol = 0;
wchar_t *p = & string [0];
for (;;) {
wchar_t kar = *p++;
if (kar == '\n' || kar == '\0') break;
if (kar == ' ' || kar == '\t') continue;
ncol ++;
do { kar = *p++; } while (kar != ' ' && kar != '\t' && kar != '\n' && kar != '\0');
if (kar == '\n' || kar == '\0') break;
}
ncol --;
if (ncol < 1) Melder_throw ("No columns.");
/*
* Count elements.
*/
p = & string [0];
nelements = 0;
for (;;) {
wchar_t kar = *p++;
if (kar == '\0') break;
if (kar == ' ' || kar == '\t' || kar == '\n') continue;
nelements ++;
do { kar = *p++; } while (kar != ' ' && kar != '\t' && kar != '\n' && kar != '\0');
if (kar == '\0') break;
}
/*
* Check if all columns are complete.
*/
if (nelements == 0 || nelements % (ncol + 1) != 0)
Melder_throw ("The number of elements (", nelements, ") is not a multiple of the number of columns plus 1 (", ncol + 1, ").");
/*
* Create empty table.
*/
nrow = nelements / (ncol + 1) - 1;
autoTableOfReal me = TableOfReal_create (nrow, ncol);
/*
* Read elements.
*/
p = & string [0];
while (*p == ' ' || *p == '\t') { Melder_assert (*p != '\0'); p ++; }
while (*p != ' ' && *p != '\t') { Melder_assert (*p != '\0'); p ++; } // ignore the header of the zeroth column ("rowLabel" perhaps)
for (long icol = 1; icol <= ncol; icol ++) {
while (*p == ' ' || *p == '\t') { Melder_assert (*p != '\0'); p ++; }
static MelderString buffer = { 0 };
MelderString_empty (& buffer);
while (*p != ' ' && *p != '\t' && *p != '\n') {
MelderString_appendCharacter (& buffer, *p);
p ++;
}
TableOfReal_setColumnLabel (me.peek(), icol, buffer.string);
MelderString_empty (& buffer);
}
for (long irow = 1; irow <= nrow; irow ++) {
while (*p == ' ' || *p == '\t' || *p == '\n') { Melder_assert (*p != '\0'); p ++; }
static MelderString buffer = { 0 };
MelderString_empty (& buffer);
while (*p != ' ' && *p != '\t') {
MelderString_appendCharacter (& buffer, *p);
p ++;
}
TableOfReal_setRowLabel (me.peek(), irow, buffer.string);
MelderString_empty (& buffer);
for (long icol = 1; icol <= ncol; icol ++) {
while (*p == ' ' || *p == '\t' || *p == '\n') { Melder_assert (*p != '\0'); p ++; }
MelderString_empty (& buffer);
while (*p != ' ' && *p != '\t' && *p != '\n' && *p != '\0') {
MelderString_appendCharacter (& buffer, *p);
p ++;
}
my data [irow] [icol] = Melder_atof (buffer.string); /* If cell contains a string, this will be 0. */
MelderString_empty (& buffer);
}
}
return me.transfer();
} catch (MelderError) {
Melder_throw ("TableOfReal: tab-separated file ", file, " not read.");
}
}
