dimAll.py (2785B)
1 import numpy as np 2 import pandas as pd 3 import decision_tree 4 from sklearn.model_selection import KFold 5 from sklearn.metrics import accuracy_score 6 7 df = pd.read_csv('../../datasets/dim.csv') 8 y = df[df.columns[-1]] 9 df = df.drop(df.columns[-1], axis=1) 10 X = df.to_numpy(dtype=np.float64) 11 12 # Define hyperparameters 13 depths = [1,2,3,4,5] 14 criteria = ["gini", "twoing", "information gain"] 15 lcs = [1,2] 16 n_splits = 5 17 n_trials = 10 18 seed = 71 19 20 # Store results 21 accuracies = {lc: {depth: [] for depth in depths} for lc in lcs} 22 treeSizes = {lc: {depth: [] for depth in depths} for lc in lcs} 23 foldSizes = [] 24 foldAccuracies = [] 25 26 # Store results 27 results = {criterion: {lc: {depth: [] for depth in depths} for lc in lcs} for criterion in criteria} 28 tree_sizes = {criterion: {lc: {depth: [] for depth in depths} for lc in lcs} for criterion in criteria} 29 30 for trial in range(n_trials): 31 kf = KFold(n_splits=n_splits, shuffle=True, random_state=(trial + seed)) 32 33 for criterion in criteria: 34 for lc in lcs: 35 for depth in depths: 36 fold_accuracies = [] 37 fold_sizes = [] 38 39 for train_index, test_index in kf.split(X): 40 X_train, X_test = X[train_index], X[test_index] 41 y_train, y_test = y[train_index], y[test_index] 42 43 clf = decision_tree.ELCClassifier(depth, lc, 100, criterion) 44 clf.fit(X_train.ravel(), y_train.shape[0], y_train, int(X_train.size / y_train.shape[0])) 45 preds = clf.predict(X_test.ravel(), y_test.shape[0], int(X_test.size / y_test.shape[0])) 46 47 fold_accuracies.append(100 * accuracy_score(y_pred=preds, y_true=y_test)) 48 fold_sizes.append(clf.getSplits() + 1) # leaves not splits 49 50 results[criterion][lc][depth].append(np.mean(fold_accuracies)) 51 tree_sizes[criterion][lc][depth].append(np.mean(fold_sizes)) 52 53 with open("resultsDim.txt", "w") as f: 54 f.write("Results:\n") 55 for criterion in criteria: 56 f.write("\n") 57 f.write(f"{criterion}:\n") 58 for lc in lcs: 59 for depth in depths: 60 avg_accuracy = np.mean(results[criterion][lc][depth]) 61 std_accuracy = np.std(results[criterion][lc][depth]) 62 avg_size = np.mean(tree_sizes[criterion][lc][depth]) 63 std_size = np.std(tree_sizes[criterion][lc][depth]) 64 65 f.write(f"LCs: {lc}, Depth: {depth} Avg Accuracy: {avg_accuracy:.1f} (Std: {std_accuracy:.1f})") 66 f.write(f", Avg # of Leaves: {avg_size:.1f} (Std: {std_size:.1f})\n")