DataMiningAnalysis1.py

#RB = Running Back
#OL = Outside Linebacker
#WR = Wide Receiver
#TE = Tight End
#FB = Full Back
#DB = Defensive Back
#QB = Quarter Back
#LB = Linebacker
#DL = Defensive Lineman

def middle(L):
    L = sorted(L)
    n = len(L)
    m = n - 1
    return (L[n/2] + L[m/2]) / 2.0

def main():
    masterDict = {} # will only be added to, not reset
    #with open('tester.txt') as f:
    with open('DataSet_DEADNFLPLAYERS.txt') as f:
        for line in f:
            current_line = line.split(',')
            #print current_line
            masterDict.update({len(masterDict) : current_line})

    #print masterDict
    total_lifeSpan = 0
    listAges = []
    for key, value in masterDict.items():
        if(key != 0):

            deathYear = int(float(value[10].strip())) #10th element
            birthYear = int(float(value[len(value)-1].strip())) #last element
            age = deathYear - birthYear # determines the age at death
            listAges.append(age) # adds age to a list
            total_lifeSpan = total_lifeSpan + age # the combinded age of all players at death

    #-------------------------------------------------------------------------

    #mean, average age across all players
    print "MEAN"
    average_age = total_lifeSpan / (len(masterDict) -1)
    print average_age
    print "----------"

    #mode, the most commonly occuring value
    print "MODE"
    from collections import Counter
    data = Counter(listAges)
    print data.most_common(1)
    print "----------"

    #median, the least extreme value or the middle value
    print "MEDIAN"
    print middle(listAges)
    print "----------"


main()