def matrix_filter(self, bp_matrix):
bp_matrix.matrix_filter = self.name
file_name = "%s_%s_%s.pickle" % (bp_matrix.matrix_filter, bp_matrix.algorithm, bp_matrix.orientation)
file_name = os.path.join(constants.raw_matrix_folder, file_name)
if not self.force_recalculate and pickleutil.has_cache(file_name):
return pickleutil.get_cache(file_name)
matrix = bp_matrix.matrix
for i in xrange(len(matrix)):
row = matrix[i]
entries = 0
total = 0
for j in xrange(len(row)):
val = matrix[i][j]
if val != -1:
total += val
entries += 1
average = self.get_average(total, entries)
for j in xrange(len(row)):
val = matrix[i][j]
if val == -1:
matrix[i][j] = average
pickleutil.cache_matrix(bp_matrix, file_name)
return bp_matrix
def get_batter_matrix(self):
if not self.force_recalculate and pickleutil.has_cache(self.batter_file_name):
return pickleutil.get_cache(self.batter_file_name)
global event_data
if event_data is None:
event_data = session.query(Event).all()
batters = session.query(Batter).all()
pitchers = session.query(Pitcher).all()
matrix_data = numpy.zeros((len(batters), len(pitchers)))
for i in xrange(0, len(batters)):
batter = batters[i]
batter_events = filter(lambda x: x.batter == batter.mlb_id, event_data)
for j in xrange(0, len(pitchers)):
pitcher = pitchers[j]
sub_score = self.score_batter(batter_events, pitcher.mlb_id)
matrix_data[i][j] = sub_score
raw_matrix = BatterPitcherMatrix(matrix_data, batters, pitchers, self.base_name,
BatterPitcherMatrix.BATTER_ORIENTATION)
pickleutil.cache_matrix(raw_matrix, self.batter_file_name)
return raw_matrix
def __get_pickles(self):
folder = constants.raw_matrix_folder
pickle_files = [f for f in os.listdir(folder) if os.path.isfile(os.path.join(folder, f))]
pickle_files = filter(lambda x: "average" in x, pickle_files)
pickle_files = [os.path.join(folder, f) for f in pickle_files]
return [pickleutil.get_cache(x) for x in pickle_files]