K-Nearest Neighbors

• Instant classifier
• Store Data in KNN
• Classifying a Query Point
• Calculating Classification Probability
• Retrieving Nearest Neighbors
• Configuring Distance Metric

Instant classifier

Instant Classifier Function

Function Signature

def instant_classifier(self, x_vals, y_vals, query_point, p=3, k=2):

Parameters

• x_vals: The input data points.

• y_vals (must be converted to int32): The labels corresponding to the input data points.

• query_point: The point for which classification is to be determined.

• p (int, default=3): The power parameter for the Minkowski distance metric.

• k (int, default=2): The number of nearest neighbors to consider for classification.

Return Value

Returns the classification result for the query_point based on the k nearest neighbors in the x_vals dataset.

Description

The instant_classifier function classifies a given query_point based on the k nearest neighbors in the x_vals dataset.  The labels of the k nearest neighbors are then used to determine the classification of the query_point.

Examples

from deeprai.models import KNN

# Sample data
x_vals = [[1, 2], [2, 3], [3, 4]]
y_vals = [0, 1, 0]
query_point = [2, 2]

# Create an instance of the classifier
classifier = KNN()

# Classify the query_point
result = classifier.instant_classifier(x_vals, y_vals, query_point, p=3, k=2)
print(result)  # This will print the classification result for the query_point

Note: Ensure that y_vals is converted to int32 before passing it to the function.

Store Data in KNN

Storing Values in KNN Classifier

Function Signature

def store_vals(self, x_values, y_values, p=3, k=2):

Parameters

• x_values: The input data points to be stored in the classifier.

• y_values (must be converted to int32): The labels corresponding to the input data points to be stored in the classifier.

• p (int, default=3): The power parameter for the Minkowski distance metric, to be stored for future use.

• k (int, default=2): The number of nearest neighbors to consider for classification, to be stored for future use.

Return Value

This function does not return anything. It modifies the KNN instance by storing the provided values.

Description

The store_vals function stores the provided data points, labels, power parameter, and number of neighbors in the KNN classifier instance. This allows for the classifier to use these values in subsequent classification tasks without needing them to be provided again.

Examples

from deeprai.models import KNN

# Sample data
x_vals = [[1, 2], [2, 3], [3, 4]]
y_vals = [0, 1, 0]

# Create an instance of the classifier
classifier = KNN()

# Store the values in the classifier
classifier.store_vals(x_vals, y_vals, p=3, k=2)

Note: Ensure that y_vals is converted to int32 before storing.

Classifying a Query Point

Classifying a Query Point with KNN

Function Signature

def classify(self, query_point):

Parameters

• query_point: The point for which classification is to be determined.

Return Value

Returns the classification result for the query_point based on the stored values in the KNN instance.

Description

The classify function classifies a given query_point based on the stored values in the KNN instance. The distance between the points is calculated using the Minkowski distance metric with the stored power parameter. The labels of the stored data points are then used to determine the classification of the query_point.

It's important to note that the store_vals function must be called prior to using the classify function to ensure that the necessary values are stored in the KNN instance.

Examples

from deeprai.models import KNN

# Sample data
x_vals = [[1, 2], [2, 3], [3, 4]]
y_vals = [0, 1, 0]
query_point = [2, 2]

# Create an instance of the classifier
classifier = KNN()

# Store the values in the classifier
classifier.store_vals(x_vals, y_vals, p=3, k=2)

# Classify the query_point
result = classifier.classify(query_point)
print(result)  # This will print the classification result for the query_point

Calculating Classification Probability

Calculating Classification Probability with KNN

Function Signature

def classify_probability(self, query_point, expected_val):

Parameters

• query_point: The point for which classification probability is to be determined.

• expected_val: The label value for which the probability is to be calculated.

Return Value

Returns the probability (in percentage) that the query_point belongs to the class specified by expected_val based on the stored values in the KNN instance.

Description

The classify_probability function calculates the probability that a given query_point belongs to the class specified by expected_val. It first retrieves the nearest neighbors of the query_point using the classify_neighbors function. It then counts how many of these neighbors have the label expected_val and calculates the probability based on this count.

It's important to note that the store_vals function must be called prior to using the classify_probability function to ensure that the necessary values are stored in the KNN instance.

Examples

from deeprai.models import KNN

# Sample data
x_vals = [[1, 2], [2, 3], [3, 4]]
y_vals = [0, 1, 0]
query_point = [2, 2]

# Create an instance of the classifier
classifier = KNN()

# Store the values in the classifier
classifier.store_vals(x_vals, y_vals, p=3, k=2)

# Calculate the probability that the query_point belongs to class 1
probability = classifier.classify_probability(query_point, 1)
print(f"The probability that the query point belongs to class 1 is {probability}%")

Retrieving Nearest Neighbors

Retrieving Nearest Neighbors with KNN

Function Signature

def classify_neighbors(self, query_point):

Parameters

• query_point: The point for which the nearest neighbors are to be determined.

Return Value

Returns the indices of the k nearest neighbors to the query_point based on the stored values in the KNN instance.

Description

The classify_neighbors function retrieves the indices of the k nearest neighbors for a given query_point based on the stored values in the KNN instance.

It's important to note that the store_vals function must be called prior to using the classify_neighbors function to ensure that the necessary values are stored in the KNN instance.

Examples

from deeprai.models import KNN

# Sample data
x_vals = [[1, 2], [2, 3], [3, 4]]
y_vals = [0, 1, 0]
query_point = [2, 2]

# Create an instance of the classifier
classifier = KNN()

# Store the values in the classifier
classifier.store_vals(x_vals, y_vals, p=3, k=2)

# Retrieve the nearest neighbors of the query_point
neighbors = classifier.classify_neighbors(query_point)
print(f"The indices of the nearest neighbors to the query point are: {neighbors}")

Configuring Distance Metric

Configuring Distance Metric for KNN

Function Signature

def config_distance(self, distance):

Parameters

• distance: The name of the distance metric to be configured for the KNN instance.

Return Value

This function does not return anything. It modifies the KNN instance.

Description

The config_distance function sets the distance metric for the KNN instance.

The valid distance metrics that can be passed to this function are:

• "hamming distance"
• "minkowski distance"
• "manhattan distance"
• "euclidean distance"

Examples

from deeprai.models import KNN

# Create an instance of the classifier
classifier = KNN()

# Configure the distance metric to be used
classifier.config_distance("euclidean distance")