void test_Iterate_UserBased_Predictor()
{
clock_t begin,end;
string train_data = "E:/data/resys/corpus/1/train/base";
string test_data = "E:/data/resys/corpus/1/test/u1.test";
FileDataReader train_reader = FileDataReader(train_data);
FileDataReader test_reader = FileDataReader(test_data);
int n_user, n_item, n_rating;
n_user = 943;
n_item = 1682;
n_rating = 80000;
UserModel trainModel(n_user,n_item,n_rating);
UserModel testModel(n_user,n_item,20000);
UserModel resultModel(n_user,n_item,20000);
if(train_reader.readData(trainModel) && test_reader.readData(testModel)) {
int min_com = 0;
double min_sim = 0.00001;
int shrink_parameter = 25;
Similarity_Shrinking shrinker(min_com,min_sim,shrink_parameter);
Pearson_Similarity similarity(shrinker);
UserBased_Predictor predictor = UserBased_Predictor(30,0);
begin = clock();
predictor.train(trainModel);
end = clock();
cout << "Train time: " << (end - begin) * 1.0 / CLOCKS_PER_SEC << endl;
begin = end;
predictor.predictAll(trainModel,testModel,resultModel);
end = clock();
cout << "Predict time: " << (end - begin) * 1.0 / CLOCKS_PER_SEC << endl;
double rmse = evl_rmse(testModel, resultModel);
cout << "RMSE: " << rmse << endl;
}
}
int main(int argc,char** argv){
if(3!=argc){
cerr<<argv[0]<<" confPath confFile"<<endl;
return -1;
}
string confPath = string(argv[1]);
string confFile = string(argv[2]);
trainModel(confPath,confFile);
return 0;
}
void printUserSim()
{
clock_t begin,end;
string train_data = "E:/data/resys/corpus/1/train/base";
FileDataReader train_reader = FileDataReader(train_data);
int n_user, n_item, n_rating;
n_user = 943;
n_item = 1682;
n_rating = 80000;
UserModel trainModel(n_user,n_item,n_rating);
if(train_reader.readData(trainModel)) {
int min_com = 0;
double min_sim = 0.0;
int shrink_parameter = 30;
int k_neighbors = 30;
Similarity_Shrinking shrinker(min_com,min_sim,shrink_parameter);
Pearson_Similarity similarity(shrinker);
NeighborCollection *neighbor_cls;
neighbor_cls = new NeighborCollection[trainModel.n_user+trainModel.user_0];
similarity.similarity(trainModel, neighbor_cls);
string sim_file = "E:/data/resys/corpus/1/user_sims.txt";
ofstream fout(sim_file.c_str());
vector<Neighbor>::iterator nb_it;
for(int u = trainModel.user_0; u < trainModel.user_0 + trainModel.n_user; u++) {
double sum = 0;
int count = 0;
for(nb_it = neighbor_cls[u].begin(); nb_it != neighbor_cls[u].end(); nb_it++) {
int n = nb_it->neighbor;
double sim = nb_it->similarity;
sum += sim;
count ++;
if(count >= k_neighbors)
break;
}
count = 0;
if(sum > 0) {
for(nb_it = neighbor_cls[u].begin(); nb_it != neighbor_cls[u].end(); nb_it++) {
int n = nb_it->neighbor;
double sim = nb_it->similarity;
sim = sim/sum;
fout << u << " " << n << " " << sim << endl;
count ++;
if(count >= k_neighbors)
break;
}
}
}
fout.close();
}
}
void ActGraph::trainModel(const std::string featureFile)
{
// features
cv::Mat stat; // mean_x, mean_y, var_x, var_y
cv::Mat shapeXY; // track shapes
cv::Mat mbhX; // MBHx
cv::Mat mbhY; // MBHy
// import features from file
readFeatures(featureFile, stat, shapeXY, mbhX, mbhY);
// GMM-clustering
trainModel(stat, shapeXY, mbhX, mbhY);
}
void test_ItemBased_Predictor()
{
clock_t begin,end;
string train_data = "E:/data/resys/corpus/1/train/base";
string test_data = "E:/data/resys/corpus/1/test/u1.test";
FileDataReader train_reader = FileDataReader(train_data);
FileDataReader test_reader = FileDataReader(test_data);
int n_user, n_item, n_rating;
n_user = 943;
n_item = 1682;
n_rating = 80000;
ItemModel trainModel(n_user,n_item,n_rating);
ItemModel testModel(n_user,n_item,20000);
ItemModel resultModel(n_user,n_item,20000);
double *userRanks = new double[n_user+1];
string filepath = "E:/data/resys/corpus/1/user_com_ranks.txt";
train_reader.readUserComRank(userRanks,filepath);
if(train_reader.readData(trainModel) && test_reader.readData(testModel)) {
int min_com = 0;
double min_sim = 0.0;
int shrink_parameter = 30;
Similarity_Shrinking shrinker(min_com,min_sim,shrink_parameter);
Pearson_Similarity similarity(shrinker);
ItemBased_Predictor predictor = ItemBased_Predictor(similarity,30,0);
begin = clock();
//predictor.train(trainModel);
predictor.train(trainModel,userRanks);
end = clock();
cout << "Train time: " << (end - begin) * 1.0 / CLOCKS_PER_SEC << endl;
begin = end;
predictor.predictAll(trainModel,testModel,resultModel);
end = clock();
cout << "Predict time: " << (end - begin) * 1.0 / CLOCKS_PER_SEC << endl;
double rmse = evl_rmse(testModel, resultModel);
cout << "RMSE: " << rmse << endl;
double mae = evl_mae(testModel, resultModel);
cout << "MAE: " << mae << endl;
}
}
void trainingMode(){
/* ********************* Interface part **********************/
/* ********************* **** ************** **** **********************/
cout << endl << endl;
cout << " ================== Training mode =============" << endl;
cout << " == Enter number of training examples: "; cin >> numTrainingExamples;
cout << " == Enter number of LBFGS iterations: " ; cin >> numLbfgsIterations;
cout << " == Enter lambda: " ; cin >> lambda;
/* Checking if input is valid */
if (numTrainingExamples > 40000) numTrainingExamples = 40000;
if (numTrainingExamples <= 0) numTrainingExamples = 150;
if (numLbfgsIterations < 1) numLbfgsIterations = 5;
/* Reading model */
//NNModel* model = readModel();
ModelLoader l;
NNModel* model = l.loadModel("modelL56.nn");
trainModel(model);
}
void ProbabilityModel::train()
{
//zliczanie ilości rekordów o danej klasie
float rowCount = 0.0f;
for (Data::iterator rowIt = trainSet->begin(); rowIt != trainSet->end() ; rowIt++)
{
DataRow &row = *rowIt;
if (labelProbs.find(row.getLabel()) == labelProbs.end())
labelProbs[row.getLabel()] = 1.0f;
else
labelProbs[row.getLabel()] += 1.0f;
rowCount += 1.0f;
}
for (StrFloatMap::iterator it=labelProbs.begin() ; it != labelProbs.end(); it++)
{
(*it).second = (*it).second / rowCount;
//std::cout << (*it).second << "\n";
}
trainModel();
}
bool KMeans::train_(MatrixFloat &data){
trained = false;
if( numClusters == 0 ){
errorLog << "train_(MatrixFloat &data) - Failed to train model. NumClusters is zero!" << std::endl;
return false;
}
if( data.getNumRows() == 0 || data.getNumCols() == 0 ){
errorLog << "train_(MatrixFloat &data) - The number of rows or columns in the data is zero!" << std::endl;
return false;
}
numTrainingSamples = data.getNumRows();
numInputDimensions = data.getNumCols();
clusters.resize(numClusters,numInputDimensions);
assign.resize(numTrainingSamples);
count.resize(numClusters);
//Randomly pick k data points as the starting clusters
Random random;
Vector< UINT > randIndexs(numTrainingSamples);
for(UINT i=0; i<numTrainingSamples; i++) randIndexs[i] = i;
std::random_shuffle(randIndexs.begin(), randIndexs.end());
//Copy the clusters
for(UINT k=0; k<numClusters; k++){
for(UINT j=0; j<numInputDimensions; j++){
clusters[k][j] = data[ randIndexs[k] ][j];
}
}
return trainModel( data );
}
