def csv(self):
import csv
sep = self._init.get('CSV', 'separator')
if not sep:
print('give CSV separator in pyquan.ini file')
sys.exit(2)
CSV = csv.CSV(sep=sep)
return CSV
def process_csv_file(self, file_defn):
from data_importer.importers import csv
kls = csv.CSV(file_defn=file_defn,
options=self.options,
stdout=self.stdout,
stderr=self.stderr,
debug_mode=self.debug_mode,
meta_args=self.meta_args)
return kls.process()
class MyConfig:
conf = csv.CSV('config.csv')
@classmethod
def get_region(cls, cur_time: MyTime):
raw_record = cls.__parse_record(
cls.conf.binary_search(
lambda record: cls.__parse_record(record)['time'] > cur_time))
# TODO: None 的情况
if raw_record['time'] < cur_time:
cls.conf.cur_record() # 跳过当前记录
while True:
if not cls.conf.cur_record(move_to_next=False):
return [cls.__parse_record(cls.conf.prev_record()), None]
if cls.__parse_record(
cls.conf.cur_record())['time'] > cur_time:
cls.conf.prev_record()
break
cls.conf.prev_record()
elif raw_record['time'] > cur_time:
while True:
if cls.conf.get_cur_record_id() == 0:
return [None, cls.__parse_record(cls.conf.cur_record())]
if cls.__parse_record(
cls.conf.prev_record())['time'] <= cur_time:
break
return [
cls.__parse_record(cls.conf.cur_record()),
cls.__parse_record(cls.conf.cur_record())
]
@staticmethod
def __parse_record(record):
return {
'time': MyTime([int(item) for item in record[:-2]] + [0]),
'angle': {
'pitch': float(record[5]),
'yaw': int(record[4])
} if record else None
}
def modWeights(self):
for layer in self.layers:
for l, error in zip(layer, self.errorList):
newWeights = []
#print l.weights
for w, e in zip(l.weights, error):
new = w - (l.learn * e * l.actv)
newWeights.append(new)
l.weights = newWeights
#print l.weights
#getting data and user preferences
data = csv.CSV().data[:10]
np.seterr(all='ignore')
numLayers = int(raw_input('how many hidden layers?: ')) + 1
nodeCounts = []
for x in range(numLayers):
if x == numLayers - 1:
nodeCounts.append(int(raw_input('how many targets: ')))
else:
nodeCounts.append(
int(raw_input('nodes for hidden layer ' + str(x + 1) + ': ')) +
1) #add 1 for bias
########################
#initializing neural network
n = Network(len(data[0]), numLayers, nodeCounts)
return fileName[fIndex]
################################
def multiply(data, yes):
if yes:
newData = []
for x in range(5):
for d in data:
newData.append(d)
return newData
return data
p = prompt()
data = csv.CSV(p[0], p[1]).data
data = multiply(data, p[1])
d = DecisionTree(data, p[2])
d.swap()
d.fit()
d.predict()
d.compare()
show(d)
d.display(d.nodes.children[0])
d.display(d.nodes.children[len(d.nodes.children) - 1], 2)
################################
###############################
def prompt():
fileName = {
'1': ('parties.txt', False, True),
'2': ('iris.txt', False, False),
'3': ('lenses.txt', True, False),
'4': ('credit.txt', True, False)
}
for key, value in fileName.iteritems():
print key + ': ' + value[0]
fIndex = ''
while not fIndex in fileName:
fIndex = raw_input('Which file#: ')
return fileName[fIndex]
################################
p = prompt()
d = DecisionTree(csv.CSV(p[0], p[1]).data, p[2])
d.start()
show(d)
################################