add pca to swig for geodawb

swig/build.sh

@@ -0,0 +1,7 @@
+
+cp OGRLayerProxy.cpp ../ShapeOperations/OGRLayerProxy.cpp
+
+swig -python -c++ -threads proxy.i
+
+CFLAGS='-Wall -O0 -DDEBUG' python setup_centos.py build_ext --inplace --force
+

swig/geoda.py

@@ -96,6 +96,53 @@ def __iter__(self): return self
 SwigPyIterator_swigregister = _geoda.SwigPyIterator_swigregister
 SwigPyIterator_swigregister(SwigPyIterator)
 
+class VecFloat(_object):
+    __swig_setmethods__ = {}
+    __setattr__ = lambda self, name, value: _swig_setattr(self, VecFloat, name, value)
+    __swig_getmethods__ = {}
+    __getattr__ = lambda self, name: _swig_getattr(self, VecFloat, name)
+    __repr__ = _swig_repr
+    def iterator(self): return _geoda.VecFloat_iterator(self)
+    def __iter__(self): return self.iterator()
+    def __nonzero__(self): return _geoda.VecFloat___nonzero__(self)
+    def __bool__(self): return _geoda.VecFloat___bool__(self)
+    def __len__(self): return _geoda.VecFloat___len__(self)
+    def pop(self): return _geoda.VecFloat_pop(self)
+    def __getslice__(self, *args): return _geoda.VecFloat___getslice__(self, *args)
+    def __setslice__(self, *args): return _geoda.VecFloat___setslice__(self, *args)
+    def __delslice__(self, *args): return _geoda.VecFloat___delslice__(self, *args)
+    def __delitem__(self, *args): return _geoda.VecFloat___delitem__(self, *args)
+    def __getitem__(self, *args): return _geoda.VecFloat___getitem__(self, *args)
+    def __setitem__(self, *args): return _geoda.VecFloat___setitem__(self, *args)
+    def append(self, *args): return _geoda.VecFloat_append(self, *args)
+    def empty(self): return _geoda.VecFloat_empty(self)
+    def size(self): return _geoda.VecFloat_size(self)
+    def clear(self): return _geoda.VecFloat_clear(self)
+    def swap(self, *args): return _geoda.VecFloat_swap(self, *args)
+    def get_allocator(self): return _geoda.VecFloat_get_allocator(self)
+    def begin(self): return _geoda.VecFloat_begin(self)
+    def end(self): return _geoda.VecFloat_end(self)
+    def rbegin(self): return _geoda.VecFloat_rbegin(self)
+    def rend(self): return _geoda.VecFloat_rend(self)
+    def pop_back(self): return _geoda.VecFloat_pop_back(self)
+    def erase(self, *args): return _geoda.VecFloat_erase(self, *args)
+    def __init__(self, *args): 
+        this = _geoda.new_VecFloat(*args)
+        try: self.this.append(this)
+        except: self.this = this
+    def push_back(self, *args): return _geoda.VecFloat_push_back(self, *args)
+    def front(self): return _geoda.VecFloat_front(self)
+    def back(self): return _geoda.VecFloat_back(self)
+    def assign(self, *args): return _geoda.VecFloat_assign(self, *args)
+    def resize(self, *args): return _geoda.VecFloat_resize(self, *args)
+    def insert(self, *args): return _geoda.VecFloat_insert(self, *args)
+    def reserve(self, *args): return _geoda.VecFloat_reserve(self, *args)
+    def capacity(self): return _geoda.VecFloat_capacity(self)
+    __swig_destroy__ = _geoda.delete_VecFloat
+    __del__ = lambda self : None;
+VecFloat_swigregister = _geoda.VecFloat_swigregister
+VecFloat_swigregister(VecFloat)
+
 class VecString(_object):
     __swig_setmethods__ = {}
     __setattr__ = lambda self, name, value: _swig_setattr(self, VecString, name, value)
@@ -457,6 +504,10 @@ def Hinge15(*args):
 def Hinge30(*args):
   return _geoda.Hinge30(*args)
 Hinge30 = _geoda.Hinge30
+
+def PCA(*args):
+  return _geoda.PCA(*args)
+PCA = _geoda.PCA
 # This file is compatible with both classic and new-style classes.
 
 

swig/install.py

@@ -0,0 +1,7 @@
+
+cp OGRLayerProxy.cpp ../ShapeOperations/OGRLayerProxy.cpp
+
+rm -rf build/
+
+python setup.py install
+

swig/proxy.cpp

@@ -3,13 +3,16 @@
 #include <set>
 #include <utility>
 #include <algorithm>
+#include <sstream>
 #include <stdio.h>
 
 #include <wx/wxprec.h>
 #ifndef WX_PRECOMP
     #include <wx/wx.h>
 #endif
 
+#include <wx/tokenzr.h>
+
 #include "../ShapeOperations/GwtWeight.h"
 #include "../ShapeOperations/GalWeight.h"
 #include "../ShapeOperations/PolysToContigWeights.h"
@@ -19,6 +22,7 @@
 #include "../Explore/CatClassification.h"
 #include "../SpatialIndAlgs.h"
 #include "../GenUtils.h"
+#include "../pca.h"
 
 #include "proxy.h"
 
@@ -513,3 +517,130 @@ Hinge30(
 ) {
     return Hinge1530(1, num_obs, data, num_categories, useScientificNotation, breaks);
 }
+
+
+///////////////////////////////////////////////////////////////////////////////
+//
+// std::vector<int> clusters
+///////////////////////////////////////////////////////////////////////////////
+std::string 
+PCA(
+    std::vector<float>& x,
+    std::vector<std::string>& x_names,
+    int _nrows,
+    int _ncols,
+    int _is_corr,
+    int _is_center,
+    int _is_scale
+) {
+    Pca pca;
+
+    bool is_corr = _is_corr == 0 ? false : true;
+    bool is_center = _is_center == 0 ? false : true;
+    bool is_scale = _is_scale == 0 ? false : true;
+
+    int rtn = pca.Calculate(x, _nrows, _ncols, is_corr, is_center, is_scale);
+
+    if ( 0 != rtn ) {
+        // error of PCA
+        return "";
+    }
+
+    vector<float> sd = pca.sd();
+    vector<float> prop_of_var = pca.prop_of_var();
+    vector<float> cum_prop = pca.cum_prop();
+    vector<float> scores = pca.scores();
+
+    vector<unsigned int> el_cols = pca.eliminated_columns();
+
+    float kaiser = pca.kaiser();
+    float thresh95 = pca.thresh95();
+
+    unsigned int ncols = pca.ncols();
+    unsigned int nrows = pca.nrows();
+
+    int max_sel_name_len = 0;
+
+    wxString method = pca.method();
+
+    wxString pca_log;
+    //pca_log << "\n\nPCA method: " << method;
+
+    pca_log << "\n\nStandard deviation:\n";
+    for (int i=0; i<sd.size();i++) pca_log << sd[i] << " ";
+
+    pca_log << "\n\nProportion of variance:\n";
+    for (int i=0; i<prop_of_var.size();i++) pca_log << prop_of_var[i] << " ";
+
+    pca_log << "\n\nCumulative proportion:\n";
+    for (int i=0; i<cum_prop.size();i++) pca_log << cum_prop[i] << " ";
+
+    pca_log << "\n\nKaiser criterion: " << kaiser;
+    pca_log << "\n\n95% threshold criterion: " << thresh95;
+
+    pca_log << "\n\nEigenvalues:\n";
+    std::stringstream ss;
+    ss << pca.eigen_values;
+    pca_log << ss.str();
+
+    //pca_log << pca.eigen_values;
+    pca_log << "\n\nEigenvectors:\n";
+
+    std::stringstream ss1;
+    ss1 << pca.eigen_vectors;
+    wxString loadings =  ss1.str();
+    wxStringTokenizer tokenizer(loadings, "\n");
+    wxArrayString items;
+    bool header = false;
+    while ( tokenizer.HasMoreTokens() )
+    {
+        wxString token = tokenizer.GetNextToken();
+        // process token here
+        items.Add(token);
+
+        if (header == false) {
+            pca_log << wxString::Format("%-*s", max_sel_name_len+4, "");
+            int n_len = token.length();
+            int pos = 0;
+            bool start = false;
+            int  sub_len = 0;
+            int pc_idx = 1;
+
+            while (pos < n_len){
+                if ( start && sub_len > 0 && (token[pos] == ' ' || pos == n_len-1) ) {
+                    // end of a number
+                    pca_log << wxString::Format("%*s%d", sub_len-1, "PC", pc_idx++);
+                    sub_len = 1;
+                    start = false;
+                } else {
+                    if (!start && token[pos] != ' ') {
+                        start = true;
+                    }
+                    sub_len += 1;
+                }
+                pos += 1;
+            }
+            header = true;
+            pca_log << "\n";
+        }
+    }
+
+    for (int k=0; k<items.size();k++) {
+        pca_log << wxString::Format("%-*s", max_sel_name_len+4, wxString(x_names[k])) << items[k] << "\n";
+    }
+
+    unsigned int row_lim;
+    unsigned int col_lim; 
+
+    if (scores.size() != nrows * ncols) {
+        row_lim = (nrows < ncols)? nrows : ncols,
+        col_lim = (ncols < nrows)? ncols : nrows;
+    } else {
+        row_lim = nrows;
+        col_lim = ncols;
+    }
+
+    return string(pca_log.mb_str());
+}
+
+

swig/proxy.h

@@ -39,3 +39,14 @@ Hinge30(
     bool useScientificNotation,
     std::vector<double>& breaks // return results
 );
+
+std::string 
+PCA(
+    std::vector<float>& x,
+    std::vector<std::string>& x_names,
+    int nrows,
+    int ncols,
+    int is_corr,
+    int is_center,
+    int is_scale
+);

swig/proxy.i

@@ -4,6 +4,7 @@
 
 %include "std_vector.i"
 namespace std {
+  %template(VecFloat) vector<float>;
   %template(VecString) vector<string>;
   %template(VecDouble) vector<double>;
   %template(VecVecDouble) vector< vector<double> >;
@@ -58,3 +59,13 @@ Hinge30(
     std::vector<double>& breaks // return results
 );
 
+std::string 
+PCA(
+    std::vector<float>& x,
+    std::vector<std::string>& x_names,
+    int nrows,
+    int ncols,
+    int is_corr,
+    int is_center,
+    int is_scale
+);