make random_state sklearn compliant

xcsf-dev · Nov 3, 2023 · 3cd548d · 3cd548d
1 parent ab835aa
commit 3cd548d
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 38 deletions.
diff --git a/cfg/default.json b/cfg/default.json
@@ -1,6 +1,6 @@
 {
     "omp_num_threads": 8,
-    "random_state": 0,
+    "random_state": -1,
     "pop_init": true,
     "max_trials": 100000,
     "perf_trials": 1000,

diff --git a/xcsf/param.c b/xcsf/param.c
@@ -60,7 +60,7 @@ param_init(struct XCSF *xcsf, const int x_dim, const int y_dim,
     param_set_x_dim(xcsf, x_dim);
     param_set_y_dim(xcsf, y_dim);
     param_set_omp_num_threads(xcsf, 8);
-    param_set_random_state(xcsf, 0);
+    param_set_random_state(xcsf, -1);
     param_set_pop_init(xcsf, true);
     param_set_max_trials(xcsf, 100000);
     param_set_perf_trials(xcsf, 1000);
@@ -608,10 +608,10 @@ const char *
 param_set_random_state(struct XCSF *xcsf, const int a)
 {
     xcsf->RANDOM_STATE = a;
-    if (a > 0) {
-        rand_init_seed(a);
-    } else {
+    if (a < 0) {
         rand_init();
+    } else {
+        rand_init_seed(a);
     }
     return NULL;
 }

diff --git a/xcsf/pybind_wrapper.cpp b/xcsf/pybind_wrapper.cpp
@@ -327,36 +327,6 @@ class XCS
 
     /* Supervised learning */
 
-    /**
-     * @brief Sets XCSF input and output dimensions.
-     * @param [in] n_x_dim Number of input dimensions.
-     * @param [in] x1 Size of second input dimension.
-     * @param [in] n_y_dim Number of output dimensions.
-     * @param [in] y1 Size of second output dimension.
-     */
-    void
-    set_dims(int n_x_dim, int x1, int n_y_dim, int y1)
-    {
-        py::dict kwargs;
-        kwargs["n_actions"] = 1;
-        if (n_x_dim > 1) {
-            kwargs["x_dim"] = x1;
-        } else {
-            kwargs["x_dim"] = 1;
-        }
-        if (n_y_dim > 1) {
-            kwargs["y_dim"] = y1;
-        } else {
-            kwargs["y_dim"] = 1;
-        }
-        // update external params dict
-        for (const auto &item : kwargs) {
-            params[item.first] = item.second;
-        }
-        // flush param update to make sure neural nets resize
-        set_params(params);
-    }
-
     /**
      * @brief Loads an input data structure for fitting.
      * @param [in,out] data Input data structure used to point to the data.
@@ -853,6 +823,13 @@ class XCS
         py::object parsed_json = json.attr("loads")(json_str);
         py::dict result(parsed_json);
         params = result;
+        // map None types
+        if (params.contains("random_state")) {
+            py::object rs = params["random_state"];
+            if (py::isinstance<py::int_>(rs) && py::int_(rs) < 0) {
+                params["random_state"] = py::none();
+            }
+        }
         free(json_str);
     }
 
@@ -877,14 +854,23 @@ class XCS
     set_params(py::kwargs kwargs)
     {
         py::dict kwargs_dict(kwargs);
+        // update external params dict
+        for (const auto &item : kwargs_dict) {
+            params[item.first] = item.second;
+        }
+        // map None types
+        if (kwargs_dict.contains("random_state")) {
+            py::object rs = kwargs["random_state"];
+            if (rs.is_none()) {
+                kwargs_dict["random_state"] = -1;
+            }
+        }
+        // convert dict to JSON and parse parameters
         py::module json_module = py::module::import("json");
         py::object json_dumps = json_module.attr("dumps")(kwargs_dict);
         std::string json_str = json_dumps.cast<std::string>();
         const char *json_params = json_str.c_str();
         param_json_import(&xcs, json_params);
-        for (const auto &item : kwargs_dict) {
-            params[item.first] = item.second;
-        }
         return *this;
     }