check.py

#!/anaconda/bin/python
import sys

def GetIntersection(source_data, check_data):
    num_intersection = 0
    for row in check_data:
        if row in source_data:
            num_intersection+=1

    return num_intersection

def PrintJaccard(source_filename, check_filename):
    source_data = []
    check_data = []

    num_lines_source = 0
    num_lines_check = 0
    num_lines_intersection = 0
    union = 0

    source_file = open(source_filename, 'r')
    check_file = open(check_filename, 'r')

    for source_line in source_file:
        source_line_elements = source_line.split(',')
        if len(source_line_elements)>2:
            num_lines_source += 1
            source_data.append([float(source_line_elements[0]), #lat
                                float(source_line_elements[1]), #lng
                                float(source_line_elements[2])])#timestep

    for check_line in check_file:
        try:
            check_line_elements = check_line.split(',')
            if len(check_line_elements)>2:
                num_lines_check += 1
                check_data.append([float(check_line_elements[0]), #lat
                                    float(check_line_elements[1]), #lng
                                    float(check_line_elements[2])])#timestep
        except:
            continue

    check_file.close()
    source_file.close()

    num_lines_intersection = GetIntersection(source_data, check_data)
    union = num_lines_source+num_lines_check-num_lines_intersection

    jaccard = float(num_lines_intersection)/union

    print "%f"%jaccard

if __name__ == "__main__":
    arguments = sys.argv
    source_filename = arguments[1]
    check_filename = arguments[2]
    PrintJaccard(source_filename,check_filename)