void append_const(double const_to_add) {
int r_num = m_feat.rowNum()+1, c_num = m_feat.colNum();
m_feat.resize(r_num, c_num);
int col;
for (col = 0; col < c_num; ++col) {
AzIFarr ifa;
m_feat.col(col)->nonZero(&ifa);
ifa.put(r_num-1, const_to_add);
m_feat.col_u(col)->load(&ifa);
}
}
static void getFloats(const char *str, AzByte dlm, AzDvect *v) {
AzIFarr ifa;
AzStrPool sp;
getStrings(str, dlm, &sp);
int ix;
for (ix = 0; ix < sp.size(); ++ix) {
ifa.put(-1, atof(sp.c_str(ix)));
}
v->reform(ifa.size());
for (ix = 0; ix < ifa.size(); ++ix) {
v->set(ix, ifa.get(ix));
}
}
/*-------------------------------------------------------------*/
void AzDmat::convert(AzSmat *m_out)
{
m_out->reform(row_num, col_num);
int col;
for (col = 0; col < col_num; ++col) {
if (column[col] != NULL) {
AzIFarr ifq;
column[col]->nonZero(&ifq);
if (ifq.size() > 0) {
m_out->col_u(col)->load(&ifq);
}
}
}
}
/*------------------------------------------*/
static void apply_log(AzSmat *m_x) {
int col;
for (col = 0; col < m_x->colNum(); ++col) {
AzIFarr ifa;
const AzSvect *v_x = m_x->col(col);
AzCursor cur;
for ( ; ; ) {
double val;
int row = v_x->next(cur, val);
if (row < 0) break;
if (val > 0) val = log(val+1);
else val = -log(-val+1);
ifa.put(row, val);
}
m_x->col_u(col)->load(&ifa);
}
}
/*------------------------------------------*/
static void find_max(const AzSmat *m_x, double ratio, int topk, AzDvect *v_max) {
AzSmat m_tran;
m_x->transpose(&m_tran);
int kk = MAX(topk, (int)((double)m_x->colNum()*ratio));
v_max->reform(m_x->rowNum());
double *max = v_max->point_u();
int row;
for (row = 0; row < m_x->rowNum(); ++row) {
AzDvect v(m_tran.col(row));
v.abs();
AzIFarr ifa;
v.nonZero(&ifa);
ifa.sort_Float(false); /* descending order */
ifa.cut(kk);
max[row] = ifa.sum() / (double)kk;
}
}
/*------------------------------------------*/
static void cap(double capval, AzSmat *m_x) {
if (capval < 0) throw new AzException("AzSvDataS::cap", "cap must be non-negative");
int col;
for (col = 0; col < m_x->colNum(); ++col) {
AzIFarr ifa;
const AzSvect *v_x = m_x->col(col);
AzCursor cur;
for ( ; ; ) {
double val;
int row = v_x->next(cur, val);
if (row < 0) break;
if (val > 0) val = MIN(val, capval);
else val = MAX(val, -capval);
ifa.put(row, val);
}
m_x->col_u(col)->load(&ifa);
}
}
/*------------------------------------------------------------------*/
void AzSvDataS::_parseDataLine_Sparse(const AzByte *inp,
int inp_len,
int f_num,
const char *data_fn,
int line_no,
/*--- output ---*/
AzIFarr &ifa_ex_val)
{
const char *eyec = "AzSvDataS::_parseDataLine_Sparse";
AzIntArr ia_isUsed;
ia_isUsed.reset(f_num, 0);
int *isUsed = ia_isUsed.point_u();
const AzByte *wp = inp, *line_end = inp + inp_len;
// AzIFarr ifa_ex_val;
for ( ; ; ) {
if (wp >= line_end) break;
AzBytArr str_token;
AzTools::getString(&wp, line_end, &str_token);
if (str_token.getLen() > 0) {
int ex;
double val;
decomposeFeat(str_token.c_str(), line_no, &ex, &val);
if (ex < 0 || ex >= f_num) {
AzBytArr s("Error in line# "); s.cn(line_no);
s.c(": invalid feature# "); s.cn(ex);
throw new AzException(AzInputError, eyec, s.c_str());
}
if (isUsed[ex]) {
AzBytArr s("Error in line# "); s.cn(line_no);
s.c(": feature# "); s.cn(ex); s.c(" appears more than once.");
throw new AzException(AzInputError, eyec, s.c_str());
}
if (val != 0) {
ifa_ex_val.put(ex, val);
}
isUsed[ex] = 1;
}
}
// m_feat->load(col, &ifa_ex_val);
}
/*-------------------------------------------------------------------------*/
void AzPrepText2::set_ifeat(const AzSmat *m_feat, int top_num,
int col, int offs, AzIFarr *ifa_ctx, feat_info fi[2]) const
{
if (col < 0 || col >= m_feat->colNum()) return;
AzIFarr ifa; m_feat->col(col)->nonZero(&ifa);
fi[0].update(ifa);
if (top_num > 0 && ifa.size() > top_num) {
ifa.sort_FloatInt(false); /* descending order */
ifa.cut(top_num);
}
fi[1].update(ifa);
if (offs == 0) ifa_ctx->concat(&ifa);
else {
for (int ix = 0; ix < ifa.size(); ++ix) {
int row = ifa.getInt(ix);
double val = ifa.get(ix);
ifa_ctx->put(row+offs, val);
}
}
}
/*------------------------------------------------------------------*/
void AzSvDataS::_parseDataLine(const AzByte *inp,
int inp_len,
int f_num,
const char *data_fn,
int line_no,
/*--- output ---*/
AzIFarr &ifa_ex_val)
{
const char *eyec = "AzSvDataS::_parseDataLine";
ifa_ex_val.prepare(f_num);
const AzByte *wp = inp, *line_end = inp + inp_len;
// AzIFarr ifa_ex_val;
int ex = 0;
for ( ; ; ) {
if (wp >= line_end) break;
#if 1
int str_len;
const AzByte *str = AzTools::getString(&wp, line_end, &str_len);
if (str_len > 0) {
if (ex >= f_num) {
AzBytArr s("Error in "); s.c(data_fn); s.c(": Line#="); s.cn(line_no);
AzPrint::writeln(log_out, s);
s.reset(); s.c("Too many values per line: expected "); s.cn(f_num); s.c(" values.");
throw new AzException(AzInputNotValid, eyec, s.c_str());
}
#if 1
double val = my_atof((char *)str, eyec, line_no);
#else
double val = atof((char *)str);
if (val == 0 && *str != '0' && *str != '+' && *str != '-') {
AzBytArr s("Invalid token "); s.c(str, str_len); s.c(" in "); s.c(data_fn); s.c(": Line#="); s.cn(line_no);
AzPrint::writeln(log_out, s);
throw new AzException(AzInputNotValid, eyec, s.c_str());
}
#endif
if (val != 0) {
ifa_ex_val.put(ex, val);
}
++ex;
}
#else
AzBytArr str_token;
AzTools::getString(&wp, line_end, &str_token);
if (str_token.getLen() > 0) {
if (ex >= f_num) {
AzBytArr s("Error in "); s.c(data_fn); s.c(": Line#="); s.cn(line_no);
AzPrint::writeln(log_out, s);
s.reset(); s.c("Too many values per line: expected "); s.cn(f_num); s.c(" values.");
throw new AzException(AzInputNotValid, eyec, s.c_str());
}
/* double val = atof(str_token.c_str()); */
double val = my_atof(str_token.c_str(), eyec, line_no);
if (val != 0) {
ifa_ex_val.put(ex, val);
}
++ex;
}
#endif
}
if (ex < f_num) {
AzTimeLog::print("Error in Line#=", line_no, log_out);
throw new AzException(AzInputNotValid, eyec, "Too few values");
}
// m_feat->load(col, &ifa_ex_val);
}