nearest_neighbor1.py

import math

class NearestNeighbor(object):
    def __init__(self):
        self._known_rows = []
        
    def learn_from_row(self, row):
        self._known_rows.append(row)
        
    def predict_missing_column(
        self, row_with_missing_column, missing_column_number):
            
        # Given a row with a missing column, try to predict the value of that
        # column by using the nearest-neighbor scheme on all the rows in
        # self._known_rows.
        
        nearest_neighbor = None
        nearest_neighbor_distance = float('inf')
        
        for row in self._known_rows:
            distance_sq = 0
            for i in [x for x in range(len(row)) if x != missing_column_number]:
                a = row[i]
                b = row_with_missing_column[i]
                if ((isinstance(a, float) or isinstance(a, int))
                    and (isinstance(b, float) or isinstance(b, int))):
                    distance_sq += (a - b) ** 2
                else:
                    distance_sq += 1
            
            distance = math.sqrt(distance_sq)
            if distance < nearest_neighbor_distance:
                nearest_neighbor_distance = distance
                nearest_neighbor = row
                
        return nearest_neighbor[missing_column_number]
                
if __name__ == '__main__':
    # Simple example.
    model = NearestNeighbor()
    
    row_0 = [27.2, 'red',  1992, 0]
    row_1 = [15.1, 'blue', 1999, 1]
    row_2 = [18.0, 'red',  1989, 0]
    
    for row in [row_0, row_1, row_2]:
        model.learn_from_row(row)
        
    print model.predict_missing_column([23.0, 'blue', 2001], 3)