def __init__(self,team_1=None, team_2=None,
classifiers=None):
random.seed()
self.playcounts = {"Run":0,"Pass":0}
self.team_1 = None
self.team_2 = None
self.game_classifiers = classifiers
score = {}
score[self.team_1] = 0
score[self.team_2] = 0
self.score = score
self.game_stats = {}
self.yards_to_first_down = 10
self.yards_to_toucdown = 80
self.yardline = 80
self.down = 1
self.O = None
self.D = None
self.series_first_down = 0
self.time_left_this_quarter = 900
self.time_left_this_game = 4*900
self. quarter = 1
self.winner = None
self.switchOffense = False
self.turn_over_on_downs = False
self.scored = False
self.summary = []
self.n_neighbors = 6
self.R = data_ops.RunData('2013_nfl_pbp_through_wk_12.csv')
self.runX,self.runY = self.R.RunTrainData()
self.knn = learn.learnKNN(self.runX, self.runY,self.n_neighbors)
def getRunModel(self, fname):
R = data_ops.RunData(fname)
self.DEF_STATS = R.getOppDefenseStats(self.team_name)
self.allXdata, self.allYdata = R.RunTrainData(fname)
# self.teamRunningData = R.getRunPlays(data_ops.getReader(fname),self.team_name)
self.teamRunningX, self.teamRunningY = R.RunTrainData(
fname, self.team_name)
def EvalAllRun():
RM = data_ops.RunData('./data/2013_nfl_pbp_through_wk_12.csv')
XT, YT = RM.RunTrainData(fname='./data/2013_nfl_pbp_through_wk_12.csv')
#for i in range(len(XT)):
# YT.append(XT[i][-1])
# YT.append(XT[i].pop())
#--Evaluate KNN with yardage --#
EvalKNN(XT, YT, n=21)
print "Average error: " + str(sum(ntests) / float(len(ntests)))
def EvalTeamRun():
for tm in map(str, data_ops.rosters.keys()):
RM = data_ops.RunData('./data/2013_nfl_pbp_through_wk_12.csv')
XT, YT = RM.RunTrainData(fname='./data/2013_nfl_pbp_through_wk_12.csv',
teamname=tm)
#for i in range(len(XT)):
# YT.append(XT[i][-1])
# YT.append(XT[i].pop())
#--Evaluate KNN with yardage --#
EvalKNN(XT, YT, n=21)
print tm + ": " + str(ntests[-1])
print "Average error: " + str(sum(ntests) / float(len(ntests)))
def EvalRUNwithSVM():
p = 0.25
RM = data_ops.RunData('./data/2013_nfl_pbp_through_wk_12.csv')
XT, YT = RM.RunTrainData(fname='./data/2013_nfl_pbp_through_wk_12.csv',
teamname='SF')
xtrain, xtest, ytrain, ytest = V.train_test_split(XT,
YT,
test_size=p,
random_state=0)
oc = SVM.SVC(kernel='rbf', probability=True)
svr_rbf = SVR(kernel='rbf', C=1e3, gamma=0.1)
model = svr_rbf.fit(xtrain, ytrain)
kmodel = oc.fit(xtrain, ytrain)
y_scores = kmodel.predict(xtest)
print kmodel.score(xtest, ytest)
eval_regres(ytest, y_scores)
from sklearn.svm import SVR
import numpy as np
import learn
import team
from sklearn import linear_model
import re
import json
from sklearn import preprocessing
from sklearn.metrics import accuracy_score
import data_ops
Xall = []
teams = []
RM = data_ops.RunData('./data/2013_nfl_pbp_through_wk_12.csv')
Xa, Yalltrain = RM.RunTrainData()
#Xf = [e[0:-2] for e in Xa]
Xalltrain = []
if True:
for x in Xa:
q = x[4] / 4.0
f = x[5] / 10.0
y = x[6] / 100.0
Xalltrain.append(x[0:3] + [q] + [f] + [y] + x[7:])
else:
Xalltrain = Xa
clf = linear_model.LogisticRegression(C=1.0, penalty='l1') # tol=1e-6)
model = clf.fit(Xalltrain, Yalltrain)
from sklearn.metrics import average_precision_score