All four algos refactored.

README.md

+# music_genre_classification
\ No newline at end of file

ann.py

+from keras.models import Sequential
+from keras.layers import Dense
+from keras.optimizers import Adam
+
+def nn_fit_predict(X_train, y_train, X_test):
+    model = Sequential([
+        Dense(58, input_shape=(58,), activation='relu'),
+        Dense(58, input_shape=(58,), activation='relu'),
+        Dense(10,  activation='softmax'),
+    ])
+
+    model.compile(optimizer="Adam", loss='categorical_crossentropy', metrics=['accuracy'])
+    model.fit(X_train, y_train, batch_size=10,epochs=100)
+    return model.predict(X_test)
+  
+#   ANN_score=np.zeros(5,dtype=float)
+# i=0
+# for train_index,test_index in kf.split(X):
+#   X_train,X_test,y_train,y_test=X[train_index],X[test_index],Y[train_index],Y[test_index]
+#   ANN_score[i]=fn_ANN(X_train,X_test,y_train,y_test)
+#   i=i+1
+# avg_ANN_score=np.sum(ANN_score)/5.0
+# print(avg_ANN_score)
\ No newline at end of file

random_forest.py

@@ -11,10 +11,10 @@ import numpy as np
   
 
 def rf_fit_predict(X_train, y_train, X_test):
-  classifier = RandomForestClassifier(max_samples=0.95, n_estimators= 3000, bootstrap=True, min_samples_split=2, min_samples_leaf=1, criterion="entropy", random_state=0)
+  classifier = RandomForestClassifier(n_estimators= 100, criterion="entropy", random_state=0)
   # classifier = ExtraTreesClassifier(max_samples=0.75, n_estimators= 3000, bootstrap=True, min_samples_split=2, min_samples_leaf=1, criterion="entropy", random_state=0)
   classifier.fit(X_train,y_train)
-  # print(classifier.get_params())
+  print("Hypperparameters: ", classifier.get_params())
   return classifier.predict(X_test)
 
 def randomized_search_fold_size_rf_fit_predict(X_train, y_train, X_test):

run_modula.py

+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+from util import *
+from ann import *
+from sklearn.metrics import confusion_matrix
+from sklearn.model_selection import KFold
+
+# def load_data():
+#     dataset = pd.read_csv("data/features_30_sec.csv")
+#     X = dataset.iloc[:, 1:59].values
+#     y = dataset.iloc[:, 59].values
+
+#     #importing the dataset
+#     from sklearn.model_selection import train_test_split
+#     X_train, X_test = train_test_split(X, test_size=0.25, random_state= 0)
+#     y_train, y_test = train_test_split(y, test_size=0.25, random_state= 0)
+
+#     #splitting our data set into training set and test set
+#     from sklearn.model_selection import train_test_split
+#     X_train, X_test = train_test_split(X, test_size=0.25, random_state= 0)
+#     y_train, y_test = train_test_split(y, test_size=0.25, random_state= 0)
+
+#     #feature scaling
+#     from sklearn.preprocessing import StandardScaler
+#     sc_X = StandardScaler()
+#     X_train = sc_X.fit_transform(X_train)
+#     X_test = sc_X.transform(X_test)
+
+#     return X_train, X_test, y_train, y_test
+    
+# def get_accuracy(cm):
+#     sum = 0
+#     for i in range(cm.shape[0]):
+#         sum = sum + cm[i][i]
+#     return 100*(sum/np.sum(cm))
+
+# def fit_predict_print(fit_predict_function, X_train, y_train, X_test, y_test):
+#     y_pred = fit_predict_function(X_train, y_train, X_test)
+#     cm = confusion_matrix(y_test, y_pred)
+#     #print(cm)
+#     print(get_accuracy(cm))
+
+X_train, X_test, y_train, y_test = load_scale_xy_with_25p_split()
+X, y = load_preprocess_xy(0, True, True, True)
+
+kf=KFold(n_splits=5)
+
+for train_index,test_index in kf.split(X):
+  X_train,X_test,y_train,y_test=X[train_index],X[test_index],y[train_index],y[test_index]
+  fit_predict_print(nn_fit_predict,X_train,y_train,X_test,y_test)
+
+# y_pred = rf_fit_predict(X_train, y_train, X_test)
+# cm = confusion_matrix(y_test, y_pred)
+# #print(cm)
+# print(get_accuracy(cm))
+
+# fit_predict_print(randomized_search_cv_rf_fit_predict, X_train, y_train, X_test, y_test)
+
+#fit_predict_print(randomized_search_cv_rf_fit_predict, X_train, y_train, X_test, y_test)
+
+# fit_predict_print(rf_fit_predict, X_train, y_train, X_test, y_test)
+
+# y_pred = randomized_search_fold_size_rf_fit_predict(X_train, y_train, X_test)
+pass
+#fit_predict_print(nn_fit_predict, X_train, y_train, X_test, y_test)
+
+# cm = confusion_matrix(y_test, y_pred)
+# #print(cm)
+# print(get_accuracy(cm))
+
+
+
+# plt.matshow(cm)
+# plt.show()
+
+
+
+
+
+pass

svm.py

+from sklearn import svm
+
+def svm_fit_predict(X_train, y_train, X_test):
+  model=svm.SVC(C=2,kernel='rbf') #regularization parameter, radial basis function, 
+  model.fit(X_train,y_train)
+  return model.predict(X_test)
\ No newline at end of file

util.py

@@ -22,6 +22,31 @@ def load_scale_xy_with_25p_split():
 
     return X_train, X_test, y_train, y_test
 
+def load_preprocess_xy(split_percentage, scale_x, encode_y, dummify_y):
+    dataset = pd.read_csv("data/features_30_sec.csv")
+    X = dataset.iloc[:, 1:59].values
+    y = dataset.iloc[:, 59].values
+
+    if encode_y:
+        encoder = LabelEncoder()
+        y = encoder.fit_transform(y)
+
+    if scale_x:
+        sc_X = StandardScaler()
+        X = sc_X.fit_transform(X) #TODO should we scale test and train separately? Yes, but wont have much difference?
+
+    if dummify_y:
+        y = pd.get_dummies(y).values
+
+    if split_percentage > 0:
+        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = split_percentage, random_state = 0)
+        return X_train, X_test, y_train, y_test
+    
+    return X, y
+    
+
+
+
 def load_scale_x_encode_y():
     dataset = pd.read_csv("data/features_30_sec.csv")
     X = dataset.iloc[:, 1:59].values
@@ -42,6 +67,9 @@ def get_accuracy(cm):
 
 def fit_predict_print(fit_predict_function, X_train, y_train, X_test, y_test):
     y_pred = fit_predict_function(X_train, y_train, X_test)
+    if y_pred.dtype != y_test.dtype:
+        y_pred = np.argmax(y_pred,axis=1)
+        y_test = np.argmax(y_test,axis=1)
     cm = confusion_matrix(y_test, y_pred)
     #print(cm)
     print("Accuracy: ", get_accuracy(cm))