diegoripley/tippecanoe
1
0
Fork 0
mirror of https://github.com/felt/tippecanoe.git synced 2026-10-02 08:25:40 +02:00
Code Issues Packages Projects Releases Wiki Activity

Weight outlieriness by categoricality

Browse Source
This commit is contained in:
Erica Fischer
2023-02-10 12:21:20 -08:00
parent 82be5627f4
commit 5c681e50d9
3 changed files with 39 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
mbtiles.cpp
+15 -9
View File
@@ -908,6 +908,20 @@ std::map<std::string, layermap_entry> merge_layermaps(std::vector<std::map<std::
return out;
}
double categorical_hash(std::string const &val) {
unsigned hash = 0;
for (size_t i = 0; i < val.size(); i++) {
// https://en.wikipedia.org/wiki/Hash_function#Fibonacci_hashing
hash = hash * 2654435769 + val[i];
}
// extra multiply so even single-byte values are distributed
// around the circle rather than clumped together on one side.
hash = hash * 2654435769;
double angle = ((double) hash) / UINT_MAX * 2 * M_PI;
return angle;
}
void add_to_file_keys(std::map<std::string, type_and_string_stats> &file_keys, std::string const &attrib, type_and_string const &val) {
if (val.type == mvt_null) {
return;
@@ -944,15 +958,7 @@ void add_to_file_keys(std::map<std::string, type_and_string_stats> &file_keys, s
}
}
unsigned hash = 0;
for (size_t i = 0; i < val.string.size(); i++) {
// https://en.wikipedia.org/wiki/Hash_function#Fibonacci_hashing
hash = hash * 2654435769 + val.string[i];
}
// extra multiply so even single-byte values are distributed
// around the circle rather than clumped together on one side.
hash = hash * 2654435769;
double angle = ((double) hash) / UINT_MAX * 2 * M_PI;
double angle = categorical_hash(val.string);
double x = cos(angle);
double y = sin(angle);
mbtiles.hpp
+1
View File
@@ -93,5 +93,6 @@ std::map<std::string, layermap_entry> merge_layermaps(std::vector<std::map<std::
std::map<std::string, layermap_entry> merge_layermaps(std::vector<std::map<std::string, layermap_entry> > const &maps, bool trunc);
void add_to_file_keys(std::map<std::string, type_and_string_stats> &file_keys, std::string const &layername, type_and_string const &val);
double categorical_hash(std::string const &value);
#endif
tile.cpp
+23 -13
View File
@@ -241,7 +241,6 @@ static int metacmp(const std::vector<long long> &keys1, const std::vector<long l
}
double get_interestingness(const serial_feature *sf, const std::map<std::string, layermap_entry> *merged_layermaps, std::string const &layername) {
double interestingness = 0;
auto layer = merged_layermaps->find(layername);
if (layer == merged_layermaps->end()) {
fprintf(stderr, "Can't find layer %s\n", layername.c_str());
@@ -250,24 +249,35 @@ double get_interestingness(const serial_feature *sf, const std::map<std::string,
for (size_t i = 0; i < sf->keys.size(); i++) {
mvt_value key = retrieve_string(sf->keys[i], sf->stringpool, NULL);
mvt_value value = retrieve_string(sf->values[i], sf->stringpool, NULL);
#if 0
// this needs to find the layer before it can find the attribute
auto lm = merged_layermaps->find(key.string_value);
if (lm != merged_layermaps->end()) {
// how categorical is it?
auto const tass = layer->second.file_keys.find(key.string_value);
if (tass != layer->second.file_keys.end()) {
// mvt_value value = retrieve_string(sf->values[i], sf->stringpool, NULL);
double xd = lm->second.xsum / lm->second.count;
double yd = lm->second.ysum / lm->second.count;
double d = sqrt(xd * xd + yd * yd);
double xd = tass->second.xsum / tass->second.count;
double yd = tass->second.ysum / tass->second.count;
// This will be 0 for something that is not categorical at all,
// 1 for something that is so categorical that there is only one value,
// and above 0.3 or so for something that is more reasonably categorical.
double categoricality = sqrt(xd * xd + yd * yd);
if (categoricality != 1) {
double angle = categorical_hash(sf->stringpool + sf->values[i] + 1);
xd -= cos(angle);
yd -= sin(angle);
// This will be 0 for something that matches the single categorical value,
// almost 2 for something that is an extreme categorical outlier,
// or somewhere in between for more reasonably categorical values.
double interestingness = sqrt(xd * xd + yd * yd);
printf("%f: %s %s\n", categoricality * interestingness, key.string_value.c_str(), sf->stringpool + sf->values[i] + 1);
}
}
#endif
}
return interestingness;
return 0;
}
static mvt_value find_attribute_value(const serial_feature *sf, std::string key, const std::map<std::string, layermap_entry> *merged_layermaps, std::string const &layername) {