Added categorical variable support for model export

diff --git a/src/regression.q b/src/regression.q
index 9d0f6163f35b680dec7343e981928617d28b44b6..74befc2e77589ef85bf22c5f1f8465335888c1e0 100644 (file)
--- a/src/regression.q
+++ b/src/regression.q
@@ -472,6 +472,44 @@ subcommand_statistics (int *keywords, pspp_linreg_cache * c)
   statistics_keyword_output (reg_stats_tol, keywords[tol], c);
   statistics_keyword_output (reg_stats_selection, keywords[selection], c);
 }
+
+static void
+reg_print_categorical_encoding (FILE *fp, pspp_linreg_cache *c)
+{
+  int i;
+  size_t j;
+  struct pspp_linreg_coeff coeff;
+  union value *val;
+  
+  fprintf (fp, "%s", reg_export_categorical_encode_1);
+
+  for (i = 1; i < c->n_indeps; i++)   /* c->coeff[0] is the intercept. */
+    {
+      coeff = c->coeff[i];
+      if (coeff.v->type == ALPHA)
+       {
+         fprintf (fp, "struct pspp_reg_categorical_variable %s;\n\t", coeff.v->name);
+       }
+    }
+  for (i = 1; i < c->n_indeps; i++)
+    {
+      coeff = c->coeff[i];
+      if (coeff.v->type == ALPHA)
+       {
+         fprintf (fp, "%s.name = \"%s\";\n\t", coeff.v->name, coeff.v->name);
+         fprintf (fp, "%s.n_vals = %d;\n\t", coeff.v->name, coeff.v->obs_vals->n_categories);
+         fprintf (fp, "%s.values = {", coeff.v->name);
+         for (j = 0; j < coeff.v->obs_vals->n_categories - 1; j++)
+           {
+             val = cat_subscript_to_value ( (const size_t) j, coeff.v);
+             fprintf (fp, "\"%s\",\n\t\t", val->s);
+           }
+         val = cat_subscript_to_value ( (const size_t) j, coeff.v);
+         fprintf (fp, "\"%s\"};\n\n\t", val->s);
+       }
+    }
+}
+
 static void
 reg_print_depvars (FILE *fp, pspp_linreg_cache *c)
 {
@@ -517,6 +555,7 @@ subcommand_export (int export, pspp_linreg_cache *c)
       fprintf (fp, "%s", reg_preamble);
       fprintf (fp, "#include <string.h>\n#include <math.h>\n\n");
       reg_print_getvar (fp, c);
+      reg_print_categorical_encoding (fp, c);
       fprintf (fp, "%s", reg_export_t_quantiles_1);
       increment = 0.5 / (double) increment;
       for (i = 0; i < n_quantiles - 1; i++)

diff --git a/src/regression_export.h b/src/regression_export.h
index e7f880e331d4409f5b0a688aa6e3edca777a3456..8798027a9952e2aea9243adc112aaa11fc19576f 100644 (file)
--- a/src/regression_export.h
+++ b/src/regression_export.h
@@ -115,4 +115,22 @@ const char reg_export_prediction_interval_3[] = " + pspp_reg_variance (var_vals,
 "\n\tresult *= pspp_reg_t_quantile ((1.0 + p) / 2.0);\n\t"
 "result += pspp_reg_estimate (var_vals, var_names);\n\treturn result;\n}\n";
 
+/*
+  Change categorical values to binary vectors. The routine will use
+  an encoding in which a categorical variable with n values is mapped
+  to a vector with n-1 entries. Value 0 is mapped to the zero vector,
+  value 1 is mapped to a vector whose first entry is 1 and all others are
+  0, etc. For example, if a variable can have 'a', 'b' or 'c' as values,
+  then the value 'a' will be encoded as (0,0), 'b' as (1,0) and 'c' as
+  (0,1). If the design matrix used to create the model used a different
+  encoding, then the function pspp_reg_categorical_encode () will return
+  a vector which does not match its categorical value in the model.
+ */
+const char reg_export_categorical_encode_1[] = "struct pspp_reg_categorical_variable\n"
+"{\n\tchar * name;\n\tsize_t n_vals;\n\tchar *[] values;\n};\n\n"
+"static\ndouble * get_value_vector (char *v)\n{\n\tdouble *result;\n\t";
+
+const char reg_export_categorical_encode_2[] = "; i++)\n\t{\n\t\tif (strcmp (v, values[i]) == 0)"
+"\n\t\t{\n\t\t\tresult[i] = 1.0;\n\t\t}\n\t}\n\treturn result;\n}\n";
+
 #endif