[英]How to optimize Pandas DataFrame speed?
我有这个熊猫代码,但它很慢。 我如何优化它? 这意味着当我运行它时,大约需要4秒钟。 我在这里调用的代码是我一遍又一遍地调用的代码,它应该尽可能快,它目前不是......任何人都有想法?
self.dataframe = pd.DataFrame(columns=list(['O' ,'H' ,'L' ,'C' ,'RSI', 'Upper Band', 'Lower Band']))
BinanceHistoricalUrl = "https://api.binance.com/api/v1/klines?"
BinanceHistoricalPayload = {'symbol' : 'BTCUSDT','interval': '1m','limit': 100}
HistoricalRequestData = requests.get(url=BinanceHistoricalUrl, params=BinanceHistoricalPayload).json()
Lenght = len(HistoricalRequestData)
for i in range(Lenght):
O = HistoricalRequestData[i][1]
O = "{:.4f}".format(O)
O = float(O)
H = HistoricalRequestData[i][2]
H = "{:.4f}".format(H)
H = float(H)
L = HistoricalRequestData[i][3]
L = "{:.4f}".format(H)
L = float(L)
C = HistoricalRequestData[i][4]
C = "{:.4f}".format(C)
C = float(C)
# Volume = HistoricalRequestData[0]["priceData"][i]['volume']
# Volume = "{:.4f}".format(Volume)
# Volume = float(Volume)
self.dataframe = self.dataframe.append({'O': O, 'H' : H, 'L' : L, 'C' : C}, ignore_index=True)
make_RSI(self.dataframe)
make_bollinger_bands(self.dataframe)
RSI = self.dataframe['RSI'][99]
RSI = float(RSI)
UppBoll = self.dataframe['Upper Band'][99]
UndBoll = self.dataframe['Lower Band'][99]
previouscloseprice = self.dataframe['C'][99]
MA = self.dataframe['20 Day MA'][99]
DistanceUppBoll = UppBoll - MA
DistanceUppBoll = float(DistanceUppBoll)
DistanceUndBoll = UndBoll - MA
DistanceUndBoll = float(DistanceUndBoll)
self.dataframe = self.dataframe.iloc[0:0]
def make_RSI(dataframe):
delta = dataframe['C'].diff()
dUp, dDown = delta.copy(), delta.copy()
dUp[dUp < 0] = 0
dDown[dDown > 0] = 0
RolUp = dUp.rolling(14).mean()
RolDown = dDown.rolling(14).mean().abs()
RS = RolUp / RolDown
dataframe['RSI'] = 100 - (100/(1+RS))
def make_bollinger_bands(dataframe):
dataframe['20 Day MA'] = dataframe['C'].rolling(window=20).mean()
dataframe['20 Day STD'] = dataframe['C'].rolling(window=20).std()
dataframe['Upper Band'] = dataframe['20 Day MA'] + (dataframe['20 Day STD'] * 2)
dataframe['Lower Band'] = dataframe['20 Day MA'] - (dataframe['20 Day STD'] * 2)
您的代码不是真正可重现的。 来点菜吧
# first import libraries
import pandas as pd
import requests
#define functions
def make_RSI(dataframe):
delta = dataframe['C'].diff()
dUp, dDown = delta.copy(), delta.copy()
dUp[dUp < 0] = 0
dDown[dDown > 0] = 0
RolUp = dUp.rolling(14).mean()
RolDown = dDown.rolling(14).mean().abs()
RS = RolUp / RolDown
dataframe['RSI'] = 100 - (100/(1+RS))
def make_bollinger_bands(dataframe):
dataframe['20 Day MA'] = dataframe['C'].rolling(window=20).mean()
dataframe['20 Day STD'] = dataframe['C'].rolling(window=20).std()
dataframe['Upper Band'] = dataframe['20 Day MA'] + (dataframe['20 Day STD'] * 2)
dataframe['Lower Band'] = dataframe['20 Day MA'] - (dataframe['20 Day STD'] * 2)
#############
# your code #
############
BinanceHistoricalUrl = "https://api.binance.com/api/v1/klines?"
BinanceHistoricalPayload = {'symbol' : 'BTCUSDT','interval': '1m','limit': 100}
#get data
HistoricalRequestData = requests.get(url=BinanceHistoricalUrl,
params=BinanceHistoricalPayload)\
.json()
# put on a dataframe
dataframe = pd.DataFrame(HistoricalRequestData)
# consider only columns from 1 to 4(included)
dataframe = dataframe[dataframe.columns[1:5]]
# assign column names
dataframe.columns = ["O", "H", "L", "C"]
# set type float
dataframe = dataframe.astype("float64")
# call functions
make_RSI(dataframe)
make_bollinger_bands(dataframe)
目前还不是很清楚你想在最后实现什么,但是你只是使用了dataframe
的最后一行,所以你可能会考虑
last = dataframe.iloc[-1]
DistanceUppBoll = last["Upper Band"] - last["20 Day MA"]
DistanceUndBoll = last["Lower Band"] - last["20 Day MA"]
我的笔记本电脑上花了717 ms
。 我想这主要取决于你的连接速度。
注意:这里的要点是你应该尽可能避免循环。
更新:如果您尝试基于基本技术分析实施交易策略,您应该看看如何计算流媒体中的MA
。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.