amansharma2910

acc = accuracy(y_test, test_predictions)
print('Model accuracy (Scratch) = ', acc*100)

amansharma2910

def accuracy(y_test, y_preds):
    """Calculates inference accuracy of the model.
    
    Args-
        y_test- Original target labels of the test set
        y_preds- Predicted target lables
    Returns-
        acc
    """
    total_correct = 0

amansharma2910

# running inference on the test data
test_predictions = KNNClassifier(X_train, y_train, X_test, k = 5)
test_predictions

amansharma2910

def KNNClassifier(train_features, train_target, test_features, k = 5):
    """Performs KNN classification on the test feature set.
    Args- 
        train_features- This denotes the feature set of the training data
        train_target- Target lables of the training data
        test_features- Feature set of the test data; assumed unlabeled
        k (default = 5)- Number of closest neighboring training data points to be considered
    Returns-
        predictions- Array of target predictions for each test data instance 
    """

amansharma2910

dist_array = distance_from_all_training(X_test[0])
dist_array

amansharma2910

def distance_from_all_training(test_point):
    """Calculates euclidean distance of test point from all the points in the training dataset

    Args- 
        test_point- Data point from test set
    Returns- 
        dist_array- Array holding distance values for all training data points 
    """
    dist_array = np.array([])
    for train_point in X_train:

amansharma2910

def euclidean_dist(pointA, pointB):
    """Calculates the euclidean distance between two vectors (numpy arrays).

    Args-
        pointA- First vector
        pointB- Second vector
    Returns-
        distance- Euclidean distance between A and B
    """
    distance = np.square(pointA - pointB) # (ai-bi)**2 for every point in the vectors

amansharma2910

# running split function on our dataset
X_train, y_train, X_test, y_test = train_test_split(features_scaled, target, test_size = 0.2)

# printing the shape of the splits
X_train.shape, y_train.shape, X_test.shape, y_test.shape

amansharma2910

def train_test_split(features, target, test_size = 0.2):
    """Splits dataset into training and testing sets.
    
    Args- 
        features- The feature matrix of the dataset (numpy array)
        target- The target array (numpy array)
        test_size- Size of test dataset that you want
    
    Returns-
        train_features, train_target, test_features, test_target 

amansharma2910

sns.countplot('quality', data = wine_df)