Skip to content

Commit

Permalink
Change in the code when saving the velocity_subnet.csv with the hiera…
Browse files Browse the repository at this point in the history 
…rchically filtered subnetwork. Take away the intersection output and complement output too. Enriched documentation of the code
  • Loading branch information
@albertoamaduzzi
albertoamaduzzi committed Jun 25, 2024
1 parent 144a032 commit 833fbb5
Showing 1 changed file with 81 additions and 77 deletions.
158 changes: 81 additions & 77 deletions source/data_analysis.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1444,24 +1444,25 @@ void make_polygons_analysis(config &config_,std::vector<centers_fcm_base> &cente
return;
}
//----------------------------------------------------------------------------------------------------
/*
make_multimodality description:
Input:
traj: vector of traj_base (contains informations useful in input:
- average_inst_speed,v_min,v_max,sinuosity
informations about output:
- means_class (int of the class the trajectory belongs to), means_p (vector float of probability size of num_classes = config.num_tm) )
config: Configuration file
centers_fcm: vector of size num_classes = config.num_tm
Description:
Goal:
Assign each traj the values of means_class (According to the Fuzzy kmean) and means_p
NOTE:
IMPORTANT -> The classes are in order of velcoity in cw (vector of centers_fcm) from the slowest to quickest .
It is important as going on with the algorithm, you clusterize the network by the distinction of the trajectories and order will
remain.
/**
* @brief make_multimodality description: Uses the FCM algorithm to clusterize the trajectories.
* Assigns to each traj the values of 'means_class' (According to the Fuzzy kmean) and 'means_p' (probability of belonging to the assigned class)
*
* @param traj: vector of traj_base (contains informations useful in input:
* - average_inst_speed,v_min,v_max,sinuosity
* informations about output:
* - means_class (int of the class the trajectory belongs to), means_p (vector float of probability size of num_classes = config.num_tm) )
* @param config: Configuration file
* @param centers_fcm: vector of size num_classes = config.num_tm
* \details The algorithm creates a matrix 'features_data' [NumberOfTrajectories x NumberOfFeatures].
*
*
*
*
* NOTE:
* IMPORTANT -> The classes are in order of velcoity in cw (vector of centers_fcm) from the slowest to quickest .
* It is important as going on with the algorithm, you clusterize the network by the distinction of the trajectories and order will
* remain.
*/
void make_multimodality(std::vector<traj_base> &traj,config &config_,std::vector<centers_fcm_base> &centers_fcm)
{
Expand Down Expand Up @@ -1683,18 +1684,13 @@ void make_multimodality(std::vector<traj_base> &traj,config &config_,std::vector
for(auto &p:traj[n].p_cluster){
p = p/total_prob_with_all_classes;
}
// DEBUG probability normalized: CHECK
// std::cout << "Probability all classes: " << total_prob_with_all_classes << std::endl;

idx_2fcm++;
}
else{
traj[n].p_cluster.insert(it,0.0);
}

// for (auto &p : traj[n].p_cluster){
// std::cout <<"Prob Associated: "<< p << std::endl;
// }

}

Expand All @@ -1706,7 +1702,6 @@ void make_multimodality(std::vector<traj_base> &traj,config &config_,std::vector
int index_cluster = 0;
int index_chosen = 0;
// DA CONTROLLARE !!!!!!!! CHECK
// std::cout << "Trajectory number: " << n << std::endl;
if(traj[n].means_class!=10 || traj[n].means_class!=11){
for(auto p: traj[n].p_cluster){
if(p>max_probability){
Expand All @@ -1719,43 +1714,6 @@ void make_multimodality(std::vector<traj_base> &traj,config &config_,std::vector
traj[n].means_class = index_chosen;
traj[n].means_p = max_probability;
}
// DEBUG
// std::cout << "Velocity: " << traj[n].average_speed << std::endl;
// std::cout << "Class: " << traj[n].means_class << std::endl;
// std::cout << "control output before pressing a key to go on: " << std::endl;
// int number;
// std::cin >> number;

/*
if (traj[n].means_class == slow_id2)
{
traj[n].p_cluster[slow_id2] = (*(fcm2->get_membership()))(idx_2fcm, slow_id2);
//traj[n].p_cluster.push_back((*(fcm2->get_membership()))(idx_2fcm, medium_id2));
if ((*(fcm2->get_membership()))(idx_2fcm, slow_id2) > (*(fcm2->get_membership()))(idx_2fcm,medium_id2))
{
traj[n].means_class = fcm2->get_reordered_map_centers_value(slow_id2);
traj[n].means_p = traj[n].p_cluster[fcm2->get_reordered_map_centers_value(slow_id2)];
out_fcm2 << " chosen class: " << traj[n].means_class << " probability: " << traj[n].means_p << " velocity: " << centers_fcm[traj[n].means_class].feat_vector[0] <<std::endl;
}
else
{
// traj[n].means_class = num_tm;
// traj[n].means_p = traj[n].p_cluster[num_tm];
traj[n].means_class = fcm2->get_reordered_map_centers_value(medium_id2);
traj[n].means_p = traj[n].p_cluster[fcm2->get_reordered_map_centers_value(medium_id2)];
out_fcm2 << " chosen class: " << traj[n].means_class << " probability: " << traj[n].means_p << " velocity: " << centers_fcm[traj[n].means_class].feat_vector[0] <<std::endl;
}
idx_2fcm++;
}
//ADDED ALBI ORDERING CENTERS_FCM
else{
traj[n].means_class += 1;
traj[n].means_p = traj[n].p_cluster[traj[n].means_class];
out_fcm2 << " chosen class: " << traj[n].means_class << " probability: " << traj[n].means_p << " velocity: " << centers_fcm[traj[n].means_class].feat_vector[0] <<std::endl;
}
//ADDED ALBI ORDERING CENTERS_FCM
*/
out_fcm2<< " trajectory number: "<< n <<std::endl;
for(auto &p:traj[n].p_cluster){
out_fcm2 << p << ";";
Expand Down Expand Up @@ -1832,7 +1790,7 @@ void make_multimodality(std::vector<traj_base> &traj,config &config_,std::vector
// ALBI

/**
* Calculates the intersection of two vectors v1 and v2 and stores the result in vector v3.
* @brief Calculates the intersection of two vectors v1 and v2 and stores the result in vector v3.
*
* @param v1 the first vector
* @param v2 the second vector
Expand Down Expand Up @@ -1884,15 +1842,31 @@ void subnet_complementary(std::vector<int> v1, std::vector<int> intersection, st
}
}
}
/* Description selecttraj_from_vectorpolysubnet_velsubnet:
* Generates a vector of trajectories that are contained by the subnet whose label is save_label
/** @brief compute_velocity_and_time_percorrence:
* @param poly_subnet: vector of integers representing the subnet
* @param poly: vector of integers representing all the road network
* @param traj: vector of trajectories
* @param label_save: string for saving the subnet velocity and time percorrence
* NOTE: Complementary to assign_new_class where traj.new_class_means is associated according to the number of times the trajectory is found in subnet of index new_class_means
* NOTE: Here I count the trajectories that pass through the subnet
* NOTE: Launched for each different poly_subnet separately
* \details
* Takes a subnet at a time (subnets80,hierarchically_selected_subnets)
* Given the poly of the subnet
*
*
*
* @return vector of trajectories that pass through the subnetwork
* @return _velocity_subnet.csv: file containing the average speed of the trajectories that pass through the
*/

std::vector<traj_base> selecttraj_from_vectorpolysubnet_velsubnet(std::vector<int> poly_subnet, std::vector<traj_base> &traj, std::vector<poly_base> &poly, std::string label_save)
std::vector<traj_base> compute_velocity_and_time_percorrence(std::vector<int> poly_subnet, std::vector<traj_base> &traj, std::vector<poly_base> &poly, std::string label_save)
{
std::cout << "case: " << label_save << " initial number of trajectories analyzed: " << traj.size() << " number of polies subnet: " << poly_subnet.size() << " total number polies: " << poly.size() << std::endl;
// bin time
int time_step = config_.bin_time * 60; // bin_time
int num_bin = int((config_.end_time - config_.start_time) / time_step);
// Trajectories that are in the subnet
std::vector<traj_base> traj_subnet;
// INITIALIZE time2av_vel,time2velocities,time2timepercorrence {bin_0: 0.0, bin_1: 0.0, ...} , {bin_0: vector<double> 0., bin_1: vector<double> 0., ...} ,{bin_0: 0.0, bin_1: 0.0, ...}
for (auto pol : poly_subnet)
Expand All @@ -1907,8 +1881,9 @@ std::vector<traj_base> selecttraj_from_vectorpolysubnet_velsubnet(std::vector<in
}
// collect data of trajectories of interest
std::cout << "number of recognized trajectories: " << traj_subnet.size() << std::endl;
// For each trajectory
for (auto &t : traj)
{
{ // Polies the trajectory passes through
std::vector<int> found_polies;
for (auto &sp : t.path)
{ // for(auto &sp:t.stop_point){
Expand Down Expand Up @@ -1974,8 +1949,20 @@ std::vector<traj_base> selecttraj_from_vectorpolysubnet_velsubnet(std::vector<in
velocity_subnet(poly_subnet, poly, time_step, num_bin, label_save);
return traj_subnet;
}
/* Description velocity_subnet:
/** @brief velocity_subnet: Generate the file _{save_label}velocity_subnet.csv
* @param poly_subnet: vector of integers representing the subnet
* @param poly: vector of integers representing all the road network
* @param time_step: time step (config.bin_time*60)
* @param num_bin: number of bins
* @param save_label: string for saving the subnet velocity and time percorrence
* \details
*
*
*
*
*
*
*
*/

void velocity_subnet(std::vector<int> poly_subnet, std::vector<poly_base> &poly, int time_step, int num_bin, std::string save_label)
Expand Down Expand Up @@ -2053,7 +2040,7 @@ std::vector<int> simplifies_c_intersect(std::map<std::string, std::vector<int>>
spaced_file_subnet_complete_intersection.close();
std::vector<traj_base> traj_complete_intersection;
std::string label_save = "complete_intersection_";
traj_complete_intersection = selecttraj_from_vectorpolysubnet_velsubnet(net_complete_intersection, traj, poly, label_save);
traj_complete_intersection = compute_velocity_and_time_percorrence(net_complete_intersection, traj, poly, label_save);

return net_complete_intersection;
}
Expand Down Expand Up @@ -2221,6 +2208,10 @@ void assign_new_class(std::vector<traj_base> &traj,std::map<std::string, std::ve
newclassification.close();
}

/**
*
* @brief simplifies_c_complement:
*/
std::map<std::string, std::vector<int>> simplifies_c_complement(std::map<std::string, std::vector<int>> subnets80,std::vector<traj_base> &traj,std::vector<poly_base> &poly)
{
std::cout << "Compute the complete complement" << std::endl;
Expand Down Expand Up @@ -2254,7 +2245,7 @@ std::map<std::string, std::vector<int>> simplifies_c_complement(std::map<std::st
// Take the complete complementary and calculate the velocity of polies. (Look at the population and time of these data.)
std::vector<traj_base> traj_complete_complement;
std::string label_save = "complete_complement_" + k->first + "_";
traj_complete_complement = selecttraj_from_vectorpolysubnet_velsubnet(complete_complement[key1], traj, poly, label_save);
traj_complete_complement = compute_velocity_and_time_percorrence(complete_complement[key1], traj, poly, label_save);
// complete_complement[key1].clear();
}
return complete_complement;
Expand All @@ -2264,13 +2255,18 @@ std::map<std::string, std::vector<int>> simplifies_c_complement(std::map<std::st
*
* @param traj: vector of all trajectories (those initialized by make_traj)
* @param poly: vector of polies (those initialized by make_poly)
* @param subnets: map of subnets (those initialized by make_subnets) (load_subnets if already computed)
* @param subnets: map of subnets (those initialized by make_subnets) (load_subnets if already computed) in main_city-pro.cpp
* \details
* - Fill subnets80 with subnet[sub_80]
* - Compute std::map<Class (0,1,..),vector<int> polies> 'hierarchically_selected_subnets' [Contains the subnetworks such that the lower velocity does not intersect with the higher velocity subnetworks]
* - Assign new_means_class to each trajectory following the principle of biggest presence in the hierarchically selected subnetworks (computing assign_new_class)
* - Compute the velocity and time percorrence of the trajectories that live in the subnetworks (compute_velocity_and_time_percorrence)
*
*
*
* ITERATED FOR EACH CLASS: fcm_centers.size()-1
* 2- simplifies_c_intersect (a)-> subnet_intersecton (b)-> selecttraj_from_vectorpolysubnet_velsubnet (c)-> velocity_subnet
* 3- simplifies_c_complement (a)-> subnet_complement (b)-> selecttraj_from_vectorpolysubnet_velsubnet (c)-> velocity_subnet
* 2- simplifies_c_intersect (a)-> subnet_intersecton (b)-> compute_velocity_and_time_percorrence (c)-> velocity_subnet
* 3- simplifies_c_complement (a)-> subnet_complement (b)-> compute_velocity_and_time_percorrence (c)-> velocity_subnet
*
*Output:
* 1 (a) -> traj_of_interest (b)-> class_i_velocity_subnet.csv -- with i = [1,...,fcm_centers.size()-1] FORMAT {id_local;time;av_speed;time_percorrence}
Expand Down Expand Up @@ -2311,9 +2307,16 @@ void analysis_subnets(std::vector<traj_base> &traj,std::vector<poly_base> &poly,
std::map<std::string,std::vector<int>> hierarchically_selected_subnets;
hierarchically_selected_subnets = hierarchical_deletion_of_intersection(subnets80);
assign_new_class(traj,hierarchically_selected_subnets);
hierarchically_selected_subnets.clear();
// VELOCITY TIME SUBNET FOR EACH CLASS
std::cout << "Print all subnets? " << config_.all_subnets_speed << " (1 = yes, 0 = no)" << std::endl;
for (auto &hssn : hierarchically_selected_subnets)
{
std::vector<traj_base> traj_of_interest;
std::string label_save = "class_" + hssn.first ;
traj_of_interest = compute_velocity_and_time_percorrence(hssn.second, traj, poly, label_save);
traj_of_interest.clear();
}
hierarchically_selected_subnets.clear();
/*
for (std::map<std::string, std::vector<int>>::iterator i = subnets.begin(); i != subnets.end(); ++i){
std::vector<traj_base> traj_subnet_tmp;
std::vector<std::string> tokens;
Expand All @@ -2331,7 +2334,7 @@ void analysis_subnets(std::vector<traj_base> &traj,std::vector<poly_base> &poly,
spaced_file_subnet << p << " ";
}
spaced_file_subnet.close();
traj_subnet_tmp = selecttraj_from_vectorpolysubnet_velsubnet(polies, traj, poly, "class_"+tokens[1]);
traj_subnet_tmp = compute_velocity_and_time_percorrence(polies, traj, poly, "class_"+tokens[1]);
}
else
Expand All @@ -2342,6 +2345,7 @@ void analysis_subnets(std::vector<traj_base> &traj,std::vector<poly_base> &poly,
traj_subnet_tmp.clear();
}
// COMPUTE INTERSECTION // PRODUCE VELOCITY AND TIME SUBNETS:
if(config_.complete_intersection_speed==true){
std::vector<int> total_intersection = simplifies_c_intersect(subnets80,traj,poly);
Expand All @@ -2352,7 +2356,7 @@ void analysis_subnets(std::vector<traj_base> &traj,std::vector<poly_base> &poly,
std::map<std::string, std::vector<int>> total_complement = simplifies_c_complement(subnets80,traj,poly);
total_complement.clear();
}

*/
}
/* Description dump_FD:
*/
Expand Down

0 comments on commit 833fbb5

Please sign in to comment.