void statistic_queries(char *input_file, int *query, int l)
{
char *endptr;
char *idx, *val, *label;
double y;
FILE *fp = fopen(input_file,"r");
for(int i=0;i<l;i++)
{
readline(fp);
label = strtok(line," \t\n");
if(label == NULL)
exit_input_error(i+1);
y = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
exit_input_error(i+1);
if(!strcmp(idx,"qid"))
{
errno = 0;
query[i] = (int) strtol(val, &endptr, 10);
if(endptr == val || errno !=0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
}
}
rewind(fp);
fclose(fp);
}
problem Flexible_vector::Read_file_without_index(std::string filename)
{
std::cout<<"IN Read_file_without_index"<<std::endl;
std::ifstream file_in(filename.c_str());
std::string e,line;
problem prob;
int elements, max_index, inst_max_index;
char *endptr;
prob.l = 0;
elements = 0;
max_index = 0;
while (std::getline(file_in, line))
{
//std::cout<<std::endl;
inst_max_index = 0;
std::stringstream lineStream(line);
std::getline(lineStream, e,','); //label
if(e.empty()) //empty line
exit_input_error(prob.l+1);
prob.y.push_back(stringToNum<double>(e));
//std::cout<<prob.y[prob.y.size()-1]<<',';
//prob.y.push_back(atof(e));
//if(endptr == e || *endptr != '\0')
// exit_input_error(prob.l+1);
prob.x_ptr.push_back(elements);
//feature
while(std::getline(lineStream, e,','))
{
inst_max_index++;
node x_tmp;
x_tmp.index = inst_max_index;
x_tmp.value = stringToNum<double>(e);
//x_tmp.value = atof(e);
prob.x.push_back(x_tmp);
//std::cout<<prob.x[prob.x.size()-1].value<<',';
elements++;
}
prob.x_interval.push_back(inst_max_index);
if(inst_max_index > max_index)
max_index = inst_max_index;
prob.l++;
}
prob.max_feature = max_index;
return prob;
}
void predict(FILE *input, FILE *output)
{
int correct = 0;
int total = 0;
double error = 0;
double sump = 0, sumt = 0, sumpp = 0, sumtt = 0, sumpt = 0;
int svm_type=svm_get_svm_type(model);
int nr_class=svm_get_nr_class(model);
double *prob_estimates=NULL;
int j;
if(predict_probability)
{
if (svm_type==NU_SVR || svm_type==EPSILON_SVR)
printf("Prob. model for test data: target value = predicted value + z,\nz: Laplace distribution e^(-|z|/sigma)/(2sigma),sigma=%g\n",svm_get_svr_probability(model));
else
{
int *labels=(int *) malloc(nr_class*sizeof(int));
svm_get_labels(model,labels);
prob_estimates = (double *) malloc(nr_class*sizeof(double));
fprintf(output,"labels");
for(j=0;j<nr_class;j++)
fprintf(output," %d",labels[j]);
fprintf(output,"\n");
free(labels);
}
}
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
label = strtok(line," \t");
target_label = strtod(label,&endptr);
if(endptr == label)
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr-1) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct svm_node *) realloc(x,max_nr_attr*sizeof(struct svm_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
++i;
}
x[i].index = -1;
if (predict_probability && (svm_type==C_SVC || svm_type==NU_SVC))
{
predict_label = svm_predict_probability(model,x,prob_estimates);
fprintf(output,"%g",predict_label);
for(j=0;j<nr_class;j++)
fprintf(output," %g",prob_estimates[j]);
fprintf(output,"\n");
}
else
{
predict_label = svm_predict(model,x);
fprintf(output,"%g\n",predict_label);
}
if(predict_label == target_label)
++correct;
error += (predict_label-target_label)*(predict_label-target_label);
sump += predict_label;
sumt += target_label;
sumpp += predict_label*predict_label;
sumtt += target_label*target_label;
sumpt += predict_label*target_label;
++total;
}
if (svm_type==NU_SVR || svm_type==EPSILON_SVR)
{
printf("Mean squared error = %g (regression)\n",error/total);
printf("Squared correlation coefficient = %g (regression)\n",
((total*sumpt-sump*sumt)*(total*sumpt-sump*sumt))/
((total*sumpp-sump*sump)*(total*sumtt-sumt*sumt))
);
}
else
printf("Accuracy = %g%% (%d/%d) (classification)\n",
(double)correct/total*100,correct,total);
if(predict_probability)
free(prob_estimates);
}
void binary_class_predict(FILE *input, FILE *output){
int total = 0;
int *labels;
int max_nr_attr = 64;
struct svm_node *x = Malloc(struct svm_node, max_nr_attr);
dvec_t dec_values;
ivec_t true_labels;
int svm_type=svm_get_svm_type(model);
if (svm_type==NU_SVR || svm_type==EPSILON_SVR){
fprintf(stderr, "wrong svm type.");
exit(1);
}
labels = Malloc(int, svm_get_nr_class(model));
svm_get_labels(model, labels);
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
label = strtok(line," \t");
target_label = strtod(label,&endptr);
if(endptr == label)
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr - 2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct svm_node *) realloc(x,max_nr_attr*sizeof(struct svm_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
++i;
}
x[i].index = -1;
predict_label = svm_predict(model,x);
fprintf(output,"%g\n",predict_label);
double dec_value;
svm_predict_values(model, x, &dec_value);
true_labels.push_back((target_label > 0)? 1: -1);
if(labels[0] <= 0) dec_value *= -1;
dec_values.push_back(dec_value);
}
// validation_function(dec_values, true_labels);
accuracy(dec_values, true_labels);
bac(dec_values, true_labels);
free(labels);
free(x);
}
void read_problem(const char *filename)
{
int elements, max_index, inst_max_index, i, j;
#ifdef _DENSE_REP
double value;
#endif
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val, *label;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob.l = 0;
elements = 0;
max_line_len = 1024;
line = Malloc(char,max_line_len);
#ifdef _DENSE_REP
max_index = 1;
while(readline(fp) != NULL)
{
char *p;
p = strrchr(line, ':');
if(p != NULL)
{
while(*p != ' ' && *p != '\t' && p > line)
p--;
if(p > line)
max_index = (int) strtol(p,&endptr,10) + 1;
}
if(max_index > elements)
elements = max_index;
++prob.l;
}
rewind(fp);
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct svm_node,prob.l);
for(i=0;i<prob.l;i++)
{
int *d;
int instance_elements = elements;
if(param.kernel_type == TRI_PRECOMPUTED ||
param.kernel_type == TRI_PRECOMPUTED_RBF)
instance_elements = (i+2);
#ifdef _FLOAT_REP
(prob.x+i)->values = Malloc(float,instance_elements);
#else
(prob.x+i)->values = Malloc(double,instance_elements);
#endif
(prob.x+i)->dim = 0;
inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
readline(fp);
label = strtok(line," \t");
prob.y[i] = strtod(label,&endptr);
if(endptr == label)
exit_input_error(i+1);
while((prob.x+i)->dim<instance_elements)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
j = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || j <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = j;
errno = 0;
value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
d = &((prob.x+i)->dim);
while (*d < j)
(prob.x+i)->values[(*d)++] = 0.0;
#ifdef _FLOAT_REP
(prob.x+i)->values[(*d)++] = (float) value;
#else
(prob.x+i)->values[(*d)++] = value;
#endif
}
}
max_index = elements-1;
#else
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
++elements;
}
++elements;
++prob.l;
}
rewind(fp);
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct svm_node *,prob.l);
x_space = Malloc(struct svm_node,elements);
max_index = 0;
j=0;
for(i=0;i<prob.l;i++)
{
inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
readline(fp);
prob.x[i] = &x_space[j];
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(i+1);
prob.y[i] = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x_space[j].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[j].index <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = x_space[j].index;
errno = 0;
#ifdef _FLOAT_REP
x_space[j].value = strtof(val,&endptr);
#else
x_space[j].value = strtod(val,&endptr);
#endif
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
}
if(inst_max_index > max_index)
max_index = inst_max_index;
x_space[j++].index = -1;
}
#endif
if(param.gamma == 0 && max_index > 0)
param.gamma = 1.0/max_index;
if(param.kernel_type == PRECOMPUTED ||
param.kernel_type == TRI_PRECOMPUTED ||
param.kernel_type == PRECOMPUTED_RBF ||
param.kernel_type == TRI_PRECOMPUTED_RBF)
for(i=0;i<prob.l;i++)
{
#ifdef _DENSE_REP
if ((prob.x+i)->dim == 0 || (prob.x+i)->values[0] == 0.0)
{
fprintf(stderr,"Wrong input format: first column must be 0:sample_serial_number\n");
exit(1);
}
if ((int)(prob.x+i)->values[0] < 0 || (int)(prob.x+i)->values[0] > max_index)
{
fprintf(stderr,"Wrong input format: sample_serial_number out of range\n");
exit(1);
}
#else
if (prob.x[i][0].index != 0)
{
fprintf(stderr,"Wrong input format: first column must be 0:sample_serial_number\n");
exit(1);
}
if ((int)prob.x[i][0].value <= 0 || (int)prob.x[i][0].value > max_index)
{
fprintf(stderr,"Wrong input format: sample_serial_number out of range\n");
exit(1);
}
#endif
}
fclose(fp);
}
void do_predict(FILE *input, FILE *output, struct model* model_)
{
int correct = 0;
int total = 0;
int nr_class=get_nr_class(model_);
double *prob_estimates=NULL;
int j, n;
int nr_feature=get_nr_feature(model_);
if(model_->bias>=0)
n=nr_feature+1;
else
n=nr_feature;
if(flag_predict_probability)
{
int *labels;
if(!check_probability_model(model_))
{
fprintf(stderr, "probability output is only supported for logistic regression\n");
exit(1);
}
labels=(int *) malloc(nr_class*sizeof(int));
get_labels(model_,labels);
prob_estimates = (double *) malloc(nr_class*sizeof(double));
fprintf(output,"labels");
for(j=0;j<nr_class;j++)
fprintf(output," %d",labels[j]);
fprintf(output,"\n");
free(labels);
}
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
int target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = 0; // strtol gives 0 if wrong format
label = strtok(line," \t");
target_label = (int) strtol(label,&endptr,10);
if(endptr == label)
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr-2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct feature_node *) realloc(x,max_nr_attr*sizeof(struct feature_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
// feature indices larger than those in training are not used
if(x[i].index <= nr_feature)
++i;
}
if(model_->bias>=0)
{
x[i].index = n;
x[i].value = model_->bias;
i++;
}
x[i].index = -1;
if(flag_predict_probability)
{
int j;
predict_label = predict_probability(model_,x,prob_estimates);
fprintf(output,"%d",predict_label);
for(j=0;j<model_->nr_class;j++)
fprintf(output," %g",prob_estimates[j]);
fprintf(output,"\n");
}
else
{
predict_label = predict(model_,x);
fprintf(output,"%d\n",predict_label);
}
if(predict_label == target_label)
++correct;
++total;
}
printf("Accuracy = %g%% (%d/%d)\n",(double) correct/total*100,correct,total);
if(flag_predict_probability)
free(prob_estimates);
}
void predict(FILE *input, FILE *output)
{
int correct = 0;
int total = 0;
double error = 0;
double sump = 0, sumt = 0, sumpp = 0, sumtt = 0, sumpt = 0;
int svm_type=svm_get_svm_type(model);
int nr_class=svm_get_nr_class(model);
double *prob_estimates=NULL;
int j;
// This block by Jianxin Wu, for average accuracy computation
int ii,label_index;
// number of correct predictions in each category
int* correct_sub = (int *)malloc(nr_class*sizeof(int));
for(ii=0;ii<nr_class;ii++) correct_sub[ii] = 0;
// number of testing examples in each category
int* total_sub = (int *)malloc(nr_class*sizeof(int));
for(ii=0;ii<nr_class;ii++) total_sub[ii] = 0;
int* labels_avg = (int*)malloc(nr_class*sizeof(int));
svm_get_labels(model,labels_avg);
if(predict_probability)
{
if (svm_type==NU_SVR || svm_type==EPSILON_SVR)
printf("Prob. model for test data: target value = predicted value + z,\nz: Laplace distribution e^(-|z|/sigma)/(2sigma),sigma=%g\n",svm_get_svr_probability(model));
else
{
int *labels=(int *) malloc(nr_class*sizeof(int));
svm_get_labels(model,labels);
prob_estimates = (double *) malloc(nr_class*sizeof(double));
fprintf(output,"labels");
for(j=0;j<nr_class;j++)
fprintf(output," %d",labels[j]);
fprintf(output,"\n");
free(labels);
}
}
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
label = strtok(line," \t");
target_label = strtod(label,&endptr);
if(endptr == label)
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr-1) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct svm_node *) realloc(x,max_nr_attr*sizeof(struct svm_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
++i;
}
x[i].index = -1;
if (predict_probability && (svm_type==C_SVC || svm_type==NU_SVC))
{
predict_label = svm_predict_probability(model,x,prob_estimates);
fprintf(output,"%g",predict_label);
for(j=0;j<nr_class;j++)
fprintf(output," %g",prob_estimates[j]);
fprintf(output,"\n");
}
else
{
predict_label = svm_predict(model,x);
fprintf(output,"%g\n",predict_label);
}
// This block by Jianxin Wu, for average accuracy
label_index = FindLabel((int)target_label,labels_avg);
total_sub[label_index]++;
if(predict_label == target_label) correct_sub[label_index]++;
if(predict_label == target_label)
++correct;
error += (predict_label-target_label)*(predict_label-target_label);
sump += predict_label;
sumt += target_label;
sumpp += predict_label*predict_label;
sumtt += target_label*target_label;
sumpt += predict_label*target_label;
++total;
}
if (svm_type==NU_SVR || svm_type==EPSILON_SVR)
{
printf("Mean squared error = %g (regression)\n",error/total);
printf("Squared correlation coefficient = %g (regression)\n",
((total*sumpt-sump*sumt)*(total*sumpt-sump*sumt))/
((total*sumpp-sump*sump)*(total*sumtt-sumt*sumt))
);
}
else
printf("Accuracy = %g%% (%d/%d) (classification)\n",
(double)correct/total*100,correct,total);
if(predict_probability)
free(prob_estimates);
// This block (till endo of function) by Jianxin WU
// Print per-category accuracy and average accuracy of categories
double sub_score = 0;
int nonempty_category = 0;
for(ii=0;ii<nr_class;ii++)
{
if(total_sub[ii]>0)
{
sub_score += (correct_sub[ii]*1.0/total_sub[ii]);
nonempty_category++;
}
}
printf("-----------\n");
for(ii=0;ii<nr_class;ii++)
{
printf("%d / %d (Category %d)\n",correct_sub[ii],total_sub[ii],labels_avg[ii]);
}
printf("-----------\n");
printf("Mean Accuray across classes = %g%%\n",sub_score*100.0/nonempty_category);
free(correct_sub);
free(total_sub);
free(labels_avg);
}
void read_problem(const char *filename)
{
int elements, max_index, inst_max_index, i, j;
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val, *label;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob.l = 0;
elements = 0;
max_line_len = 1024;
line = Malloc(char,max_line_len);
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
++elements;
}
++elements;
++prob.l;
}
rewind(fp);
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct svm_node *,prob.l);
prob.W = Malloc(double,prob.l);
x_space = Malloc(struct svm_node,elements);
max_index = 0;
j=0;
for(i=0;i<prob.l;i++)
{
inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
readline(fp);
prob.x[i] = &x_space[j];
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(i+1);
prob.y[i] = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
prob.W[i] = 1;
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x_space[j].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[j].index <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = x_space[j].index;
errno = 0;
x_space[j].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
}
if(inst_max_index > max_index)
max_index = inst_max_index;
x_space[j++].index = -1;
}
if(param.gamma == 0 && max_index > 0)
param.gamma = 1.0/max_index;
if(param.kernel_type == PRECOMPUTED)
for(i=0;i<prob.l;i++)
{
if (prob.x[i][0].index != 0)
{
fprintf(stderr,"Wrong input format: first column must be 0:sample_serial_number\n");
exit(1);
}
if ((int)prob.x[i][0].value <= 0 || (int)prob.x[i][0].value > max_index)
{
fprintf(stderr,"Wrong input format: sample_serial_number out of range\n");
exit(1);
}
}
fclose(fp);
if(weight_file)
{
fp = fopen(weight_file,"r");
for(i=0;i<prob.l;i++)
fscanf(fp,"%lf",&prob.W[i]);
fclose(fp);
}
}
// read in a problem (in libsvm format)
// clicks shows qid:1 f1:val1 f2:val2
void read_problem(const char *filename)
{
int max_index, inst_max_index, i;
long int elements, j;
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val;
//char *label;
int clicks = 0;
int shows = 0;
int lines = 0;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob.l = 0;
elements = 0;
max_line_len = 1024;
line = Malloc(char,max_line_len);
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // clicks
clicks = atoi(p);
p = strtok(NULL," \t"); // shows
shows = atoi(p);
p = strtok(NULL," \t"); // qid
if (shows <=0 || clicks > shows) {
lines++;
continue;
}
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
elements += shows;
}
elements += shows; // for bias term
prob.l += shows;
lines++;
}
rewind(fp);
/*
printf("lines:%d\n",lines);
printf("prob length:%d\n", prob.l);
printf("elements:%d\n", elements);
*/
prob.bias=bias;
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct feature_node *,prob.l);
x_space = Malloc(struct feature_node,elements+prob.l);
max_index = 0;
j = 0;
long int start = 0;
long int end = 0;
long int instances = 0;
long int feature_len = 0;
int non_clicks = 0;
int feature_node_size = sizeof(struct feature_node);
for(i=0;i<lines;i++)
{
inst_max_index = 0; // strtol gives 0 if wrong format
readline(fp);
char *p = strtok(line," \t\n");
if(p == NULL) // empty line
exit_input_error(i+1);
clicks = atoi(p);
p = strtok(NULL," \t"); // shows
shows = atoi(p);
if (shows <=0 || clicks > shows) {
continue;
}
p = strtok(NULL," \t"); // qid
start = end;
j=end;
prob.x[instances] = &x_space[j];
prob.y[instances] = 0; // label nonclicks
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x_space[j].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[j].index <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = x_space[j].index;
errno = 0;
x_space[j].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
}
if(inst_max_index > max_index)
max_index = inst_max_index;
if(prob.bias >= 0)
x_space[j++].value = prob.bias;
x_space[j++].index = -1;
end = j;
feature_len = end - start;
non_clicks = shows - clicks;
instances++;
for(int k=1; k<shows; k++) {
if (k < non_clicks) {
prob.y[instances] = 0;
} else {
prob.y[instances] = 1;
}
prob.x[instances] = &x_space[end];
memcpy(&x_space[end], &x_space[start], feature_len * feature_node_size);
/*
for(j = 0; j<feature_len;j++) {
x_space[end+j].index = x_space[start+j].index;
x_space[end+j].value = x_space[start+j].value;
}
*/
start = end;
end += feature_len;
instances++;
}
}
/*
for(i=0;i<prob.l;i++) {
printf("prob[%d]:%f\n",i,prob.y[i]);
}
for(i=0;i<elements+prob.l;i++) {
printf("x_space[%d]:%d,%f\n",i,x_space[i].index,x_space[i].value);
}
*/
if(prob.bias >= 0)
{
prob.n=max_index+1;
for(i=1;i<lines;i++)
(prob.x[i]-2)->index = prob.n;
x_space[j-2].index = prob.n;
}
else
prob.n=max_index;
fclose(fp);
}
void do_predict(FILE *input, FILE *output)
{
int total=0;
int n;
int nr_feature=get_nr_feature(model_);
double *dvec_t;
double *ivec_t;
int *query;
n=nr_feature;
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
total++;
rewind(input);
dvec_t = new double[total];
ivec_t = new double[total];
query = new int[total];
total = 0;
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = 0; // strtol gives 0 if wrong format
query[total] = 0;
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(total+1);
target_label = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(total+1);
ivec_t[total] = target_label;
while(1)
{
if(i>=max_nr_attr-2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct feature_node *) realloc(x,max_nr_attr*sizeof(struct feature_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
if (strcmp(idx,"qid") == 0)
{
errno = 0;
query[total] = (int) strtol(val,&endptr,10);
if(endptr == val || errno != 0 || *endptr != '\0')
exit_input_error(i+1);
continue;
}
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
// feature indices larger than those in training are not used
if(x[i].index <= nr_feature)
++i;
}
x[i].index = -1;
predict_label = predict(model_,x);
fprintf(output,"%.10f\n",predict_label);
dvec_t[total++] = predict_label;
}
double result[3];
eval_list(ivec_t,dvec_t,query,total,result);
info("Pairwise Accuracy = %g%%\n",result[0]*100);
info("MeanNDCG (LETOR) = %g\n",result[1]);
info("NDCG (YAHOO) = %g\n",result[2]);
}
void split(char *input_file, int l, int machines, int nr_query, struct Query_Machine *q_machine, int *query)
{
int machine_id = 0;
double y;
int len = 0;
FILE *fp = fopen(input_file,"r");
char *idx, *val, *endptr;
char *label;
char **out_file = (char**)malloc(sizeof(char*)*machines);
for(int i=0;i<machines;i++)
out_file[i] = (char*)malloc(sizeof(char)*1024);
FILE **f = (FILE**)malloc(sizeof(FILE*)*machines);
if(mkdir("temp_dir",0777)==0)
{
printf("Directory was successfully created!\n");
}
else
{
printf("Directory has existed!!\n");
}
for(int i=0;i<machines;i++)
{
sprintf(out_file[i],"temp_dir/train.txt.%d",i);
f[i] = fopen(out_file[i],"w");
}
char *copy_line = (char*)malloc(sizeof(char)*2048);
for(int j=0;j<l;j++)
{
readline(fp);
len = (int)strlen(line);
//printf("len=%d for line:%d\n",len,j+1);
if(len > 2048)
{
copy_line = (char*)realloc(copy_line,len*sizeof(char));
}
sprintf(copy_line, "%s", line);
//strcpy(copy_line,line);
//printf("copy_line:%s",copy_line);
//printf("line:%s",line);
label = strtok(line, " \t\n");
if(label == NULL)
exit_input_error(j+1);
y = strtod(label, &endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(j+1);
idx = strtok(NULL,":");
val = strtok(NULL, " \t");
if(val == NULL)
exit_input_error(j+1);
if(!strcmp(idx,"qid"))
{
errno = 0;
query[j] = (int)strtol(val, &endptr, 10);
if(endptr == val || errno !=0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(j+1);
}
for(int i=0;i<nr_query;i++)
{
if(query[j] == q_machine[i].query)
{
machine_id = q_machine[i].machine_id;
break;
}
}
fprintf(f[machine_id],"%s",copy_line);
}
free(copy_line);
for(int i=0;i<machines;i++)
free(out_file[i]);
free(out_file);
for(int i=0;i<machines;i++)
fclose(f[i]);
rewind(fp);
fclose(fp);
}
void do_predict(FILE *input, FILE *output)
{
int total = 0;
int nr_class=get_nr_class(model_);
double *prob_estimates=NULL;
int n;
int nr_feature=get_nr_feature(model_);
if(model_->bias>=0)
n=nr_feature+1;
else
n=nr_feature;
if(!check_probability_model(model_))
{
fprintf(stderr, "probability output is only supported for logistic regression\n");
exit(1);
}
prob_estimates = (double *) malloc(nr_class*sizeof(double));
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
int clicks = 0;
int shows = 0;
while(readline(input) != NULL)
{
int i = 0;
double target_ctr, predict_ctr;
char *idx, *val, *endptr;
int inst_max_index = 0; // strtol gives 0 if wrong format
char *p = strtok(line," \t\n"); //clicks
if(p == NULL) // empty line
exit_input_error(total+1);
clicks = atoi(p);
p = strtok(NULL," \t"); // shows
shows = atoi(p);
p = strtok(NULL," \t"); // qid:1
if (shows <=0 || clicks > shows) {
continue;
}
target_ctr = (double)clicks / shows;
while(1)
{
if(i>=max_nr_attr-2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct feature_node *) realloc(x,max_nr_attr*sizeof(struct feature_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
// feature indices larger than those in training are not used
if(x[i].index <= nr_feature)
++i;
}
if(model_->bias>=0)
{
x[i].index = n;
x[i].value = model_->bias;
i++;
}
x[i].index = -1;
predict_probability(model_,x,prob_estimates);
fprintf(output,"%d %d ", clicks, shows);
predict_ctr = prob_estimates[0];
fprintf(output," %g\n", predict_ctr);
}
info("total:%d\n",total);
free(prob_estimates);
}
void read_problem(const char *filename)
{
int elements[2], inst_max_index, i[2], j[2];
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val, *label;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob[0].l = 0;
prob[1].l = 0;
elements[0] = 0;
elements[1] = 0;
int current_set;
max_line_len = 1024;
line = Malloc(char,max_line_len);
while(readline(fp)!=NULL)
{
if(line[0] == '-')
current_set = 1;
else
current_set = 0;
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
++elements[current_set];
}
++elements[current_set];
++prob[current_set].l;
}
rewind(fp);
prob[0].y = Malloc(double,prob[0].l);
prob[0].x = Malloc(struct svm_node *,prob[0].l);
x_space[0] = Malloc(struct svm_node,elements[0]);
prob[1].y = Malloc(double,prob[1].l);
prob[1].x = Malloc(struct svm_node *,prob[1].l);
x_space[1] = Malloc(struct svm_node,elements[1]);
// max_index = 0;
j[0]=0;
j[1]=0;
i[0] = 0;
i[1] = 0;
while( i[0] < prob[0].l || i[1] < prob[1].l )
// for(i=0;i<prob.l;i++)
{
inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
readline(fp);
if(line[0] == '-')
current_set = 1;
else
current_set = 0;
prob[current_set].x[i[current_set]] = &x_space[current_set][j[current_set]];
label = strtok(line," \t");
prob[current_set].y[i[current_set]] = strtod(label,&endptr);
if(endptr == label)
exit_input_error(i[0]+i[1]+1);
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x_space[current_set][j[current_set]].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[current_set][j[current_set]].index <= inst_max_index)
exit_input_error(i[0]+i[1]+1);
else
inst_max_index = x_space[current_set][j[current_set]].index;
errno = 0;
x_space[current_set][j[current_set]].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i[0]+i[1]+1);
++j[current_set];
}
if(inst_max_index > max_index)
max_index = inst_max_index;
x_space[current_set][j[current_set]++].index = -1;
i[current_set]++;
}
fclose(fp);
}
/**
* This code is largely borrowed from LIBLINEAR.
* TODO: This can be made faster with SIMD.
**/
size_t create_dw_corpus(std::string filename, const size_t n_elements, const size_t n_examples,
double * & p_examples, long * & p_cols, long * & p_rows){
p_examples = new double[n_elements];
p_cols = new long[n_elements];
p_rows = new long[n_examples];
FILE *fp = fopen(filename.c_str(),"r");
ssize_t read;
size_t max_line_len = 1024;
char * line = new char[max_line_len];
char *endptr;
char *idx, *val, *label;
double * y = new double[n_examples];
int input_idx;
double input_label;
double input_value;
if(fp == NULL){
fprintf(stderr,"can't open input file %s\n",filename.c_str());
exit(1);
}
size_t maxidx = 0;
int inst_max_index, j=0;
size_t ct = 0;
for(int i=0;i<n_examples;i++){
inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
read = getline(&line, &max_line_len, fp);
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(i+1);
input_label = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
p_rows[i] = ct;
while(1){
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
input_idx = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0')
exit_input_error(i+1);
errno = 0;
input_value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
p_examples[ct] = input_value;
p_cols[ct] = input_idx;
ct ++;
//std::cout << input_idx << std::endl;
if(input_idx > maxidx){
maxidx = input_idx;
}
}
p_examples[ct] = (input_label+1)/2; // normalize +1/-1 to 1/0
p_cols[ct] = -1;
ct ++;
}
maxidx ++;
return maxidx;
}
void problem::Load(const char* filename,const NUMBER bias,const int maxdim)
{ // revised from the LIBLINEAR/LIBSVM read_problem() function
timeb TimingMilliSeconds;
ftime(&TimingMilliSeconds);
Clear();
FILE *fp = fopen(filename,"r");
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
max_line_len = 1024;
line = Malloc(char,max_line_len);
l = 0;
allocated = 0;
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
allocated++;
}
allocated++;
l++;
}
rewind(fp);
if(bias>0) allocated += l;
index_buf = new int[allocated];
value_buf = new NUMBER[allocated];
y = new int[l];
indexes = new int*[l];
values = new NUMBER*[l];
int max_index = 0;
long int j=0;
char *endptr;
char *idx, *val, *label;
for(int i=0;i<l;i++)
{
int inst_max_index = 0; // strtol gives 0 if wrong format
readline(fp);
indexes[i] = &index_buf[j];
values[i] = &value_buf[j];
label = strtok(line," \t");
y[i] = (int) strtol(label,&endptr,10);
if(endptr == label) exit_input_error(i+1);
while(1)
{
if(bias>0)
{
index_buf[j] = 1;
value_buf[j] = bias;
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
index_buf[j] = (int) strtol(idx,&endptr,10)+(bias>0);
if(endptr == idx || errno != 0 || *endptr != '\0' || index_buf[j] <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = index_buf[j];
errno = 0;
value_buf[j] = (NUMBER)strtod(val,&endptr);
if(value_buf[j]<0)
value_buf[j] = 0.001;
//else if(value_buf[j]>=1)
//value_buf[j] = 0.999;
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
}
if(inst_max_index > max_index) max_index = inst_max_index;
index_buf[j++] = -1;
}
n=max_index;
if(maxdim>0 && n>maxdim) n=maxdim;
fclose(fp);
free(line);
line = NULL;
struct timeb now;
ftime(&now);
std::cout<<"Dataset loaded in "<<int( (now.time-TimingMilliSeconds.time)*1000+(now.millitm-TimingMilliSeconds.millitm) )<<" msec."<<std::endl;
}
//struct feature_node
//{
// int index;
// double value;
//};
//void _parse_command_line(int argc, char **argv, char *input_file_name, char *model_file_name)
//#endif
//{
// int i;
// void (*print_func)(const char*) = NULL; // default printing to stdout
//
// // default values
// _param.solver_type = L2R_L2LOSS_SVC;
// _param.C = 1;
// _param.eps = INF; // see setting below
// _param.p = 0;
// _param.nr_weight = 0;
// _param.weight_label = NULL;
// _param.weight = NULL;
// _param.ite = 100; //default
// // parse options
// for(i=1;i<argc;i++)
// {
// if(argv[i][0] != '-') break;
// if(++i>=argc)
// exit_with_help();
// switch(argv[i-1][1])
// {
// case 's':
// _param.solver_type = atoi(argv[i]);
// break;
//
// case 'c':
// _param.C = atof(argv[i]);
// break;
//
// case 'p':
// _param.p = atof(argv[i]);
// break;
//
// case 'e':
// _param.eps = atof(argv[i]);
// break;
//
// case 'i':
// _param.ite = atof(argv[i]);
// break;
//
// case 'q':
// print_func = &print_null;
// i--;
// break;
//
// default:
// fprintf(stderr,"unknown option: -%c\n", argv[i-1][1]);
// exit_with_help();
// break;
// }
// }
//
// set_print_string_function(print_func);
//
// // determine filenames
// if(i>=argc)
// exit_with_help();
//
// strcpy(input_file_name, argv[i]);
//#ifdef FIGURE56
// if(i<argc-1)
// strcpy(test_file_name,argv[i+1]);
// else
// {
// exit_with_help();
// }
//#else
// if(i<argc-1)
// strcpy(model_file_name,argv[i+1]);
// else
// {
// //strrchr() 函数查找字符在指定字符串中从后面开始的第一次出现的位置，如果成功，则返回从该位置到字符串结尾的所有字符，如果失败，则返回 false。
// //与之相对应的是strchr()函数，它查找字符串中首次出现指定字符的位置。
// char *p = strrchr(argv[i],'/');
// if(p==NULL)//there are no parent directories in the path of train data, that means train data are in the current directory
// p = argv[i];
// else//the train data contain parent directoryis, ++p: the pointer move from char '/' to the first char of the input train data
// ++p;
// sprintf(model_file_name,"%s.model",p);//int sprintf ( char * str, const char * format, ... );Write formatted data to string
// }
//#endif
//
// if(_param.eps == INF)
// {
// switch(_param.solver_type)
// {
// case L2R_L2LOSS_SVC:
// _param.eps = 0.01;
// break;
// case L2R_L2LOSS_SVR:
// case WX_RBTREE:
// case Y_RBTREE:
// case AVLTREE:
// case AATREE:
// case DIRECT_COUNT:
// case SELECTION_TREE:
// case PRSVMP:
// _param.eps = 0.001;
// break;
// case L2R_L1LOSS_SVR_DUAL:
// _param.eps = 0.1;
// break;
// }
// }
//}
void read_problem(const char *filename)
{
int max_index, inst_max_index, i;
long int elements, j;
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val, *label;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob.l = 0;
elements = 0;
max_line_len = 1024;
line = Malloc(char,max_line_len);
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
elements++;
}
elements++; // for bias term
prob.l++;
}
rewind(fp);
// struct problem
// {
// int l, n;//l: total instance number(starts from 0);n:total feature number
// int *query;
// double *y;//y:label value
// struct feature_node **x;
// };
//prob.l: instances number of input file
//prob.y: label array of each instance of input file
//prob.x: pointer array of each instance's features
//prob.query:query array of each instance of input file
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct feature_node *,prob.l);
prob.query = Malloc(int,prob.l);
//x_space:apply enough spaces to store all instances' features
x_space = Malloc(struct feature_node,elements+prob.l);
max_index = 0;
j=0;
for(i=0;i<prob.l;i++)//iterate all instances
{
prob.query[i] = 0;
inst_max_index = 0; // strtol gives 0 if wrong format
readline(fp);
prob.x[i] = &x_space[j];
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(i+1);
prob.y[i] = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
if (!strcmp(idx,"qid"))//qid
{
errno = 0;
prob.query[i] = (int) strtol(val, &endptr,10);
if(endptr == val || errno !=0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
}
else//feature
{
errno = 0;
x_space[j].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[j].index <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = x_space[j].index;
errno = 0;
x_space[j].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
/*for debug
if(j==46)
printf("%d",j);*/
}
}//finish parsing one line of data
if(inst_max_index > max_index)
max_index = inst_max_index;
x_space[j++].index = -1;
}// iterate all instances
prob.n=max_index;
fclose(fp);
}
// read in a problem (in libsvm format)
void read_problem(const char *filename)
{
int max_index, inst_max_index, i;
size_t elements, j;
FILE *fp = fopen(filename,"r");
char *endptr;
char *idx, *val, *label;
if(fp == NULL)
{
fprintf(stderr,"can't open input file %s\n",filename);
exit(1);
}
prob.l = 0;
elements = 0;
max_line_len = 1024;
line = Malloc(char,max_line_len);
while(readline(fp)!=NULL)
{
char *p = strtok(line," \t"); // label
// features
while(1)
{
p = strtok(NULL," \t");
if(p == NULL || *p == '\n') // check '\n' as ' ' may be after the last feature
break;
elements++;
}
elements++; // for bias term
prob.l++;
}
rewind(fp);
prob.bias=bias;
prob.y = Malloc(double,prob.l);
prob.x = Malloc(struct feature_node *,prob.l);
x_space = Malloc(struct feature_node,elements+prob.l);
max_index = 0;
j=0;
for(i=0;i<prob.l;i++)
{
inst_max_index = 0; // strtol gives 0 if wrong format
readline(fp);
prob.x[i] = &x_space[j];
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(i+1);
prob.y[i] = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(i+1);
while(1)
{
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x_space[j].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x_space[j].index <= inst_max_index)
exit_input_error(i+1);
else
inst_max_index = x_space[j].index;
errno = 0;
x_space[j].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(i+1);
++j;
}
if(inst_max_index > max_index)
max_index = inst_max_index;
if(prob.bias >= 0)
x_space[j++].value = prob.bias;
x_space[j++].index = -1;
}
if(prob.bias >= 0)
{
prob.n=max_index+1;
for(i=1;i<prob.l;i++)
(prob.x[i]-2)->index = prob.n;
x_space[j-2].index = prob.n;
}
else
prob.n=max_index;
fclose(fp);
}
double * predict(FILE *input, int &size)
{
int correct = 0;
int total = 0;
double error = 0;
double sump = 0, sumt = 0, sumpp = 0, sumtt = 0, sumpt = 0;
int svm_type=svm_get_svm_type(model);
int nr_class=svm_get_nr_class(model);
double *prob_estimates=NULL;
size = 0;
int tmp;
while ( (tmp=fgetc(input)) != EOF) {
if (tmp == '\n')
size++;
}
rewind(input);
double * res = (double *) malloc (size*sizeof(double));
target = (double *) malloc (size*sizeof(double));
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
int j = 0;
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = -1; // strtol gives 0 if wrong format, and precomputed kernel has <index> start from 0
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(total+1);
target_label = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr-1) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct svm_node *) realloc(x,max_nr_attr*sizeof(struct svm_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
++i;
}
x[i].index = -1;
predict_label = svm_predict(model,x);
res[j] = predict_label;
target[j] = target_label;
j++;
}
return res;
}
void do_predict(FILE *input, FILE *output)
{
int correct = 0;
int total = 0;
double error = 0;
double sump = 0, sumt = 0, sumpp = 0, sumtt = 0, sumpt = 0;
int nr_class=get_nr_class(model_);
double *prob_estimates=NULL;
int j, n;
int nr_feature=get_nr_feature(model_);
if(model_->bias>=0)
n=nr_feature+1;
else
n=nr_feature;
if(flag_predict_probability)
{
int *labels;
if(!check_probability_model(model_))
{
fprintf(stderr, "probability output is only supported for logistic regression\n");
exit(1);
}
labels=(int *) malloc(nr_class*sizeof(int));
get_labels(model_,labels);
prob_estimates = (double *) malloc(nr_class*sizeof(double));
fprintf(output,"labels");
for(j=0;j<nr_class;j++)
fprintf(output," %d",labels[j]);
fprintf(output,"\n");
free(labels);
}
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = 0; // strtol gives 0 if wrong format
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(total+1);
target_label = strtod(label,&endptr);
if(endptr == label || *endptr != '\0')
exit_input_error(total+1);
while(1)
{
if(i>=max_nr_attr-2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct feature_node *) realloc(x,max_nr_attr*sizeof(struct feature_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
// feature indices larger than those in training are not used
if(x[i].index <= nr_feature)
++i;
}
if(model_->bias>=0)
{
x[i].index = n;
x[i].value = model_->bias;
i++;
}
x[i].index = -1;
if(model_->normal){
double length = 0;
for(int kk = 0; x[kk].index != -1; kk++)
length += x[kk].value * x[kk].value;
length = sqrt(length);
for(int kk = 0; x[kk].index != -1; kk++)
x[kk].value /= length;
}
if(flag_predict_probability)
{
int j;
predict_label = predict_probability(model_,x,prob_estimates);
fprintf(output,"%g",predict_label);
for(j=0;j<model_->nr_class;j++)
fprintf(output," %g",prob_estimates[j]);
fprintf(output,"\n");
}
else
{
predict_label = predict(model_,x);
fprintf(output,"%g\n",predict_label);
}
if(predict_label == target_label)
++correct;
error += (predict_label-target_label)*(predict_label-target_label);
sump += predict_label;
sumt += target_label;
sumpp += predict_label*predict_label;
sumtt += target_label*target_label;
sumpt += predict_label*target_label;
++total;
}
if(model_->param.solver_type==L2R_L2LOSS_SVR ||
model_->param.solver_type==L2R_L1LOSS_SVR_DUAL ||
model_->param.solver_type==L2R_L2LOSS_SVR_DUAL)
{
info("Mean squared error = %g (regression)\n",error/total);
info("Squared correlation coefficient = %g (regression)\n",
((total*sumpt-sump*sumt)*(total*sumpt-sump*sumt))/
((total*sumpp-sump*sump)*(total*sumtt-sumt*sumt))
);
}
else
info("Accuracy = %g%% (%d/%d)\n",(double) correct/total*100,correct,total);
if(flag_predict_probability)
free(prob_estimates);
}
void do_predict(FILE *input, FILE *output)
{
int correct = 0;
int total = 0;
double error = 0;
double sump = 0, sumt = 0, sumpp = 0, sumtt = 0, sumpt = 0;
int nr_class=get_nr_class(model_[0]);
double *prob_estimates=NULL;
int j, n;
int nr_feature=get_nr_feature(model_[0]);
if(model_[0]->bias>=0)
n=nr_feature+1;
else
n=nr_feature;
if(flag_predict_probability)
{
int *labels;
if(!check_probability_model(model_[0]))
{
fprintf(stderr, "probability output is only supported for logistic regression\n");
exit(1);
}
labels=(int *) malloc(nr_class*sizeof(int));
get_labels(model_[0],labels);
prob_estimates = (double *) malloc(nr_class*sizeof(double));
fprintf(output,"labels");
for(j=0;j<nr_class;j++)
fprintf(output," %d",labels[j]);
fprintf(output,"\n");
free(labels);
}
max_line_len = 1024;
line = (char *)malloc(max_line_len*sizeof(char));
while(readline(input) != NULL)
{
int i = 0;
double target_label, predict_label;
char *idx, *val, *label, *endptr;
int inst_max_index = 0; // strtol gives 0 if wrong format
label = strtok(line," \t\n");
if(label == NULL) // empty line
exit_input_error(total+1);
// target_label = strtod(label,&endptr);
switch (label[0]) {
case 'A': target_label = 0; break;
case 'B': target_label = 1; break;
case 'C': target_label = 1; break;
case 'D': target_label = 1; break;
}
// if(endptr == label || *endptr != '\0')
// exit_input_error(total+1);
for (int pid = 0; pid < sum_pro; pid++) {
while(1)
{
if(i>=max_nr_attr-2) // need one more for index = -1
{
max_nr_attr *= 2;
x = (struct feature_node *) realloc(x,max_nr_attr*sizeof(struct feature_node));
}
idx = strtok(NULL,":");
val = strtok(NULL," \t");
if(val == NULL)
break;
errno = 0;
x[i].index = (int) strtol(idx,&endptr,10);
if(endptr == idx || errno != 0 || *endptr != '\0' || x[i].index <= inst_max_index)
exit_input_error(total+1);
else
inst_max_index = x[i].index;
errno = 0;
x[i].value = strtod(val,&endptr);
if(endptr == val || errno != 0 || (*endptr != '\0' && !isspace(*endptr)))
exit_input_error(total+1);
// feature indices larger than those in training are not used
if(x[i].index <= nr_feature)
++i;
}
if(model_[pid]->bias>=0)
{
x[i].index = n;
x[i].value = model_[pid]->bias;
i++;
}
x[i].index = -1;
if(flag_predict_probability)
{
int j;
predict_label = predict_probability(model_[pid],x,prob_estimates);
fprintf(output,"%g",predict_label);
for(j=0;j<model_[pid]->nr_class;j++)
fprintf(output," %g",prob_estimates[j]);
fprintf(output,"\n");
}
else
{
p_label[pid] = predict(model_[pid],x);
fprintf(output,"%g", p_label[pid]);
// printf("pid%dhas done\n",pid );
}
fprintf(output, "\n" );
}
int count = 0;
predict_label = 0;
// for ( int l = 0; l < BLOCK ; l++) {
// for (int m = 0;m < BLOCK * N; m++) {
// // printf("%f\t", p_label[l * BLOCK + m]);
// if ( p_label[l * BLOCK + m] == 1) {
// // p_label[l] = 1;
// // break;
// p_label[l]++;
// // count++;* 4
// }
// }
// if (p_label[l] < 4) {
// count++;
// }
// // if ( p_label[l] == 1) {
// // predict_label = 1;
// // }
// // else {
// // predict_label = 0;
// // }
// // if (count >0) {
// // predict_label = 1;
// // }
// // else {
// // predict_label = 0;
// // }
// }
// if (count > 0 ) {
// predict_label = 0;
// }
// else {
// predict_label = 1;
// }
// /printf("\n");
// fprintf(output,"%g\n",predict_label);
// if(predict_label == target_label)
// ++correct;
// error += (predict_label-target_label)*(predict_label-target_label);
// sump += predict_label;
// sumt += target_label;
// sumpp += predict_label*predict_label;
// sumtt += target_label*target_label;
// sumpt += predict_label*target_label;
// ++total;
}
// if(check_regression_model(model_[0]))
// {
// info("Mean squared error = %g (regression)\n",error/total);
// info("Squared correlation coefficient = %g (regression)\n",
// ((total*sumpt-sump*sumt)*(total*sumpt-sump*sumt))/
// ((total*sumpp-sump*sump)*(total*sumtt-sumt*sumt))
// );
// }
// else
// info("Accuracy = %g%% (%d/%d)\n",(double) correct/total*100,correct,total);
// if(flag_predict_probability)
// free(prob_estimates);
}