def predict(self):
result = []
for code in self.data:
volume, chg_p, vma5 = Stock.get_col().index(
"volume"), Stock.get_col().index(
"chg_p"), Stock.get_col().index("vma5")
data_array = np.array(self.data[code])
try:
chg_p_array, vma5_array, volume_array = np.float32(
data_array[:, chg_p]), np.float32(
data_array[:, vma5]), np.float32(data_array[:, volume])
training_data = []
for v in volume_array:
training_data.append([v])
training_label = [["down", "up"][int(x > 5.0)]
for x in chg_p_array]
trend, percent = classify0([vma5_array[-1]],
np.array(training_data),
training_label, 5)
if trend == "up":
result.append((code, percent))
except:
trace_log()
pass
sorted_result = sorted(result,
key=operator.itemgetter(1),
reverse=True)
return sorted_result
def predict(self):
result = {}
for code in self.data:
date, close, chg_p = Stock.get_col().index("date"), Stock.get_col().index("close"), Stock.get_col().index("chg_p")
try:
last_date = self.data[code][-1][0]
if last_date != time.strftime("%Y-%m-%d", time.localtime()):
continue
except:
continue
data_array = np.array(self.data[code])
try:
close_array, chg_p_array = np.float32(data_array[:,close]), np.float32(data_array[:,chg_p])
def gold_predict(close_array):
close_array = close_array[-10:]
max_close = max(close_array)
min_close = min(close_array)
if close_array[-1] < max_close and max_close == close_array[-2]:
if (max_close - min_close)/min_close > 0.1 :
ratio = (max_close - close_array[-1]) / (max_close - min_close)
if ratio < (0.382+0.010) and ratio > (0.382-0.010):
result[code] = 2*max_close-close_array[-1]
return True
return False
top_5_close_array = close_array[-5:]
if not gold_predict(top_5_close_array):
top_10_close_array = close_array[-10:]
gold_predict(top_10_close_array)
except:
trace_log()
pass
return result.items()
def verify(self):
up5 = 0
up1 = 0
up9 = 0
up = []
chg_p = Stock.get_col().index("chg_p")
for code in self.up_code:
data_array = np.array(self.data[code])
try:
chg_p_array = np.float32(data_array[:, chg_p])
if chg_p_array[-1] >= 1.0:
up1 += 1
up.append((code, chg_p_array[-1]))
if chg_p_array[-1] >= 5.0:
up5 += 1
if chg_p_array[-1] >= 9.0:
up9 += 1
except:
logging.error("%s failed" % code)
trace_log()
print "up5 = %.2f, up1 = %.2f, up9 = %.2f" % (
float(up5) / float(len(self.up_code)), float(up1) /
float(len(self.up_code)), float(up9) / float(len(self.up_code)))
print up
def predict(self):
result = {}
for code in self.data:
volume, chg_p, vma5, ma5, close, high = Stock.get_col().index(
"volume"), Stock.get_col().index("chg_p"), Stock.get_col(
).index("vma5"), Stock.get_col().index("ma5"), Stock.get_col(
).index("close"), Stock.get_col().index("high")
try:
last_date = self.data[code][-1][0]
if last_date != time.strftime("%Y-%m-%d", time.localtime()):
continue
except:
continue
data_array = np.array(self.data[code])
try:
chg_p_array, vma5_array, volume_array, ma5_array, close_array, high_array = np.float32(
data_array[:, chg_p]), np.float32(
data_array[:, vma5]), np.float32(
data_array[:, volume]), np.float32(
data_array[:, ma5]), np.float32(
data_array[:, close]), np.float32(
data_array[:, high])
# 判断跳空
jumped = False
for i in range(-1, -10, -1):
if high_array[i] < close_array[i - 1]:
jumped = True
break
if jumped and chg_p_array[-1] > 3.0:
down_number = 0
for c in chg_p_array[-10:]:
if c < 0.0:
down_number += 1
result[code] = down_number
except:
trace_log()
pass
sorted_result = sorted(result.iteritems(),
key=operator.itemgetter(1),
reverse=True)
return sorted_result