K-Nearest Neighbors classification algorithm

Algorithm

Distance Euclidean

Classifier Majority Vote | Weighted Distance

Feature Normalization None

Dimension Reduction None

Cross-Validation None

Assumptions

• File format csv
• Input data shuffled
• Number of features < 10
• Feature units scaled
• Attribute datatype real
• Missing values false
• Headers false

Configuration

{
    "input": <path: string>,
    "exclude": <columns: integer[]>,
    "class": <column: integer>,
    "neighbors": <count: integer>,
    "skewed": <true/false: boolean>,
    "sample": <size: integer>,
    "test": <percentage: float>
}

Example

{
    "input": "../petals.csv",
    "exclude": [],
    "class": 4,
    "neighbors": 3,
    "skewed": false,
    "sample": 150,
    "test": 0.3
}