Segregate figure saving for optimization

bird/postprocess/SA_optimization/get_csv.py

@@ -31,8 +31,8 @@ def get_config_result(study_fold: str = ".") -> None:
     count = 0
 
     # Save data into CSV files
-    xfname = f"Xdata_{study_fold}.csv"
-    yfname = f"ydata_{study_fold}.csv"
+    xfname = os.path.join(study_fold, f"Xdata.csv")
+    yfname = os.path.join(study_fold, f"ydata.csv")
     with open(xfname, "w", newline="") as csvfile:
         writer = csv.writer(csvfile)
         for sims in results:
@@ -52,6 +52,10 @@ def get_config_result(study_fold: str = ".") -> None:
 
 
 if __name__ == "__main__":
-    studies = {"study_scaleup_0_4vvm_3000W": r"0.0036$m^3$ 0.4vvm 0W"}
+    studies = {
+        "study_scaleup_0_4vvm_3000W": r"608$m^3$ 0.4vvm 3000W",
+        "study_scaleup_0_1vvm_6000W": r"608$m^3$ 0.1vvm 6000W",
+        "study_0_4vvm_1W": r"0.00361$m^3$ 0.4vvm 1W",
+    }
     for study in studies:
         get_config_result(study)

bird/postprocess/SA_optimization/get_optimal.py

@@ -1,3 +1,4 @@
+import os
 import random
 import warnings
 
@@ -90,6 +91,7 @@ def run_optimization(
     n_runs: int = 10,
     max_iters: int = 1000,
     bootstrap_size: int = 100,
+    out_folder: str = ".",
 ):
     """
     Bootstraps data, runs optimization and postprocesses the results.
@@ -115,6 +117,8 @@ def run_optimization(
         maximum number of SA iterations
     bootstrap_size : int
         size of bootstrap samples
+    out_folder: str
+        folder where to output the results
 
     Returns
     ----------
@@ -175,7 +179,10 @@ def run_optimization(
         ]
     )
     df.to_csv(
-        f"best_bootstrap_solution_{model_type}_size_{bootstrap_size}.csv",
+        os.path.join(
+            out_folder,
+            f"best_bootstrap_solution_{model_type}_size_{bootstrap_size}.csv",
+        ),
         index=False,
     )
 
@@ -186,7 +193,7 @@ def run_optimization(
     upper_bound = mean_trace + 1.96 * std_trace / np.sqrt(n_runs)
     iterations = np.arange(max_iters + 1)
 
-    plt.figure(figsize=(8, 4))
+    plt.figure(figsize=(8, 6))
     plt.plot(
         iterations,
         mean_trace,
@@ -203,19 +210,25 @@ def run_optimization(
     )
     pretty_labels(
         "Iteration",
-        "Best Surrogate-Predicted Objective",
-        fontsize=16,
+        r"Predicted QOI [kg$^2$/kWh$^2$]",
+        fontsize=20,
         title=f"Mean Convergence with 95% Confidence Interval ({model_type.upper()})",
         grid=True,
         fontname="Times",
     )
-    pretty_legend(fontsize=16, fontname="Times")
+    pretty_legend(fontsize=20, fontname="Times")
     plt.savefig(
-        f"Mean_Convergence_plot_{model_type}_size_{bootstrap_size}.png",
+        os.path.join(
+            out_folder,
+            f"Mean_Convergence_plot_{model_type}_size_{bootstrap_size}.png",
+        ),
         dpi=300,
     )
     plt.savefig(
-        f"Mean_Convergence_plot_{model_type}_size_{bootstrap_size}.pdf",
+        os.path.join(
+            out_folder,
+            f"Mean_Convergence_plot_{model_type}_size_{bootstrap_size}.pdf",
+        ),
     )
     plt.show()
 
@@ -255,24 +268,56 @@ def run_optimization(
     pretty_legend(fontsize=16, fontname="Times")
     plt.tight_layout()
     plt.savefig(
-        f"Mean_L1_distance_{model_type}_size_{bootstrap_size}.png", dpi=300
+        os.path.join(
+            out_folder,
+            f"Mean_L1_distance_{model_type}_size_{bootstrap_size}.png",
+        ),
+        dpi=300,
+    )
+    plt.savefig(
+        os.path.join(
+            out_folder,
+            f"Mean_L1_distance_{model_type}_size_{bootstrap_size}.pdf",
+        )
     )
-    plt.savefig(f"Mean_L1_distance_{model_type}_size_{bootstrap_size}.pdf")
     plt.show()
 
 
 if __name__ == "__main__":
 
-    # Read data from the csv file.
-    X_raw_data = pd.read_csv("Xdata_study_scaleup_0_4vvm_3000W.csv")
-    y_raw_data = pd.read_csv("ydata_study_scaleup_0_4vvm_3000W.csv")
-
-    X = X_raw_data.values
-    y = y_raw_data.iloc[:, :-1].values
-    # We want to maximize, so we minimize the opposite
-    y = y * -1
-
-    # The function will build, tune, and optimize the surrogate model and postprocess the results.
-    run_optimization(
-        X, y, model_type="rbf", n_runs=10, max_iters=1000, bootstrap_size=150
-    )
+    # studies = ["study_scaleup_0_4vvm_3000W", "study_scaleup_0_1vvm_6000W", "study_0_4vvm_1W"]
+    # studies = ["study_scaleup_0_1vvm_6000W", "study_0_4vvm_1W"]
+    studies = ["study_scaleup_0_4vvm_3000W", "study_0_4vvm_1W"]
+
+    for study in studies:
+        # Read data from the csv file.
+        X_raw_data = pd.read_csv(os.path.join(study, "Xdata.csv"))
+        y_raw_data = pd.read_csv(os.path.join(study, "ydata.csv"))
+
+        X = X_raw_data.values
+        y = y_raw_data.iloc[:, :-1].values
+        # We want to maximize, so we minimize the opposite
+        y = y * -1
+
+        # The function will build, tune, and optimize the surrogate model and postprocess the results.
+        run_optimization(
+            X,
+            y,
+            model_type="rbf",
+            n_runs=10,
+            max_iters=1000,
+            bootstrap_size=150,
+            out_folder=study,
+        )
+        # run_optimization(
+        #    X,
+        #    y,
+        #    model_type="rf",
+        #    n_runs=10,
+        #    max_iters=1000,
+        #    bootstrap_size=150,
+        #    out_folder=study,
+        # )
+        # run_optimization(
+        #    X, y, model_type="nn", n_runs=10, max_iters=1000, bootstrap_size=150, out_folder=study
+        # )