正确将自己的函数应用于分组的熊猫数据框

Question

I have a Pandas dataframe like:

   ticket date         close  
0    AAA  2018-01-12  176.16
1    AAA  2018-01-13  176.49
3    AAA  2018-01-14  176.00
4    BBB  2018-01-12  78.19
5    BBB  2018-01-13  79.90
6    BBB  2018-01-14  78.10

I have a function:

def rsi(dataframe, period, column = 'close'):
    delta = dataframe[column].diff()
    up, down = delta.copy(), delta.copy()
    up[up < 0] = 0
    down[down > 0] = 0
    rolling_up = up.ewm(com=period - 1, adjust=False).mean()
    rolling_down = down.ewm(com= period -1, adjust=False).mean().abs()
    rsi = 100 - 100 / (1 + rolling_up / rolling_down)
    dataframe['rsi'] =  rsi
    return dataframe

What I need is to apply this function to my dataframe for each groupby('ticket'). I tried this but it doesn't work. Give me please some advice.

print(dataframe.groupby('ticket').apply(rsi, 2))

I get a error:

cannot reindex from a duplicate axis

Whole source code is:

# -*- coding: utf-8 -*-

import json
import pandas
import requests
import datetime

def get_historical_prices(tickets, range):
    request_params = {'symbols': ','.join(tickets), 'types': 'chart', 'range': range}
    json = requests.get('https://api.iextrading.com/1.0/stock/market/batch', params = request_params).json()
    united_dataframe = pandas.DataFrame()
    for symbol in json:
        ticket_dataframe = pandas.DataFrame(json[symbol]['chart'])
        ticket_dataframe.insert(0, 'ticket', symbol)
        united_dataframe = united_dataframe.append(ticket_dataframe)
    return united_dataframe[['ticket', 'date', 'close']]

def rsi(dataframe, period, column = 'close'):
    delta = all_prices[column].diff()
    up, down = delta.copy(), delta.copy()
    up[up < 0] = 0
    down[down > 0] = 0
    rolling_up = up.ewm(com=period - 1, adjust=False).mean()
    rolling_down = down.ewm(com= period -1, adjust=False).mean().abs()
    rsi = 100 - 100 / (1 + rolling_up / rolling_down)
    dataframe['rsi'] =  rsi
    return dataframe

# Get the data
tickets = ['AAPL', 'FB', 'TSLA']
all_prices = get_historical_prices(tickets, '1m')

print(all_prices.groupby('ticket').apply(rsi, 2))

Answer 1

there is a problem in the source code. the line

delta = all_prices[column].diff()

should be

delta = dataframe[column].diff() 

fixing it will also run without problem. reassignment will add the column rsi to all_prices ie

all_prices = all_prices.groupby('ticket').apply(rsi, 2)

so the final cod and results is shown below

In [20]: # -*- coding: utf-8 -*-
    ...: 
    ...: import json
    ...: import pandas
    ...: import requests
    ...: import datetime
    ...: 
    ...: def get_historical_prices(tickets, range):
    ...:     request_params = {'symbols': ','.join(tickets), 'types': 'chart', 'range': range}
    ...:     json = requests.get('https://api.iextrading.com/1.0/stock/market/batch', params = request_params).json()
    ...:     united_dataframe = pandas.DataFrame()
    ...:     for symbol in json:
    ...:         ticket_dataframe = pandas.DataFrame(json[symbol]['chart'])
    ...:         ticket_dataframe.insert(0, 'ticket', symbol)
    ...:         united_dataframe = united_dataframe.append(ticket_dataframe)
    ...:     return united_dataframe[['ticket', 'date', 'close']]
    ...: 
    ...: def rsi(dataframe, period, column = 'close'):
    ...:     delta = dataframe[column].diff()
    ...:     up, down = delta.copy(), delta.copy()
    ...:     up[up < 0] = 0
    ...:     down[down > 0] = 0
    ...:     rolling_up = up.ewm(com=period - 1, adjust=False).mean()
    ...:     rolling_down = down.ewm(com= period -1, adjust=False).mean().abs()
    ...:     rsi = 100 - 100 / (1 + rolling_up / rolling_down)
    ...:     dataframe['rsi'] = rsi
    ...:     return dataframe
    ...: 
    ...: # Get the data
    ...: tickets = ['AAPL', 'FB', 'TSLA']
    ...: all_prices = get_historical_prices(tickets, '1m')
    ...: 
    ...: all_prices = all_prices.groupby('ticket').apply(rsi, 2)
    ...: print(all_prices.head())
    ...: 
    ...: 
  ticket        date   close        rsi
0   AAPL  2018-01-12  177.09        NaN
1   AAPL  2018-01-16  176.19   0.000000
2   AAPL  2018-01-17  179.10  76.377953
3   AAPL  2018-01-18  179.26  78.208232
4   AAPL  2018-01-19  178.46  44.065484

Answer 2

The issue here is related to the line

dataframe['rsi'] =  rsi
return dataframe

the problem is that rsi does not have the same index as dataframe, further more rsi has a different length

I changed the lines above to

return rsi

and the code ran without issues

so the final cod and results is shown below

In [12]: # -*- coding: utf-8 -*-
    ...: 
    ...: import json
    ...: import pandas
    ...: import requests
    ...: import datetime
    ...: 
    ...: def get_historical_prices(tickets, range):
    ...:     request_params = {'symbols': ','.join(tickets), 'types': 'chart', 'range': range}
    ...:     json = requests.get('https://api.iextrading.com/1.0/stock/market/batch', params = request_params).json()
    ...:     united_dataframe = pandas.DataFrame()
    ...:     for symbol in json:
    ...:         ticket_dataframe = pandas.DataFrame(json[symbol]['chart'])
    ...:         ticket_dataframe.insert(0, 'ticket', symbol)
    ...:         united_dataframe = united_dataframe.append(ticket_dataframe)
    ...:     return united_dataframe[['ticket', 'date', 'close']]
    ...: 
    ...: def rsi(dataframe, period, column = 'close'):
    ...:     delta = all_prices[column].diff()
    ...:     up, down = delta.copy(), delta.copy()
    ...:     up[up < 0] = 0
    ...:     down[down > 0] = 0
    ...:     rolling_up = up.ewm(com=period - 1, adjust=False).mean()
    ...:     rolling_down = down.ewm(com= period -1, adjust=False).mean().abs()
    ...:     rsi = 100 - 100 / (1 + rolling_up / rolling_down)
    ...:     
    ...:     return rsi
    ...: 
    ...: # Get the data
    ...: tickets = ['AAPL', 'FB', 'TSLA']
    ...: all_prices = get_historical_prices(tickets, '1m')
    ...: 
    ...: print(all_prices.groupby('ticket').apply(rsi, 2))
    ...: 
    ...: 
close   0    1          2          3          4          5          6   \
ticket                                                                   
AAPL   NaN  0.0  76.377953  78.208232  44.065484  16.991057  19.694656   
FB     NaN  0.0  76.377953  78.208232  44.065484  16.991057  19.694656   
TSLA   NaN  0.0  76.377953  78.208232  44.065484  16.991057  19.694656   

close         7         8        9     ...             11         12  \
ticket                                 ...                             
AAPL    3.521704  1.252711  15.2917    ...      76.523444  48.103572   
FB      3.521704  1.252711  15.2917    ...      76.523444  48.103572   
TSLA    3.521704  1.252711  15.2917    ...      76.523444  48.103572   

close          13         14         15       16         17         18  \
ticket                                                                   
AAPL    80.777497  46.146025  23.884925  8.34273  16.897125  75.919398   
FB      80.777497  46.146025  23.884925  8.34273  16.897125  75.919398   
TSLA    80.777497  46.146025  23.884925  8.34273  16.897125  75.919398   

close          19         20  
ticket                        
AAPL    15.705838  12.501725  
FB      15.705838  12.501725  
TSLA    15.705838  12.501725  

[3 rows x 61 columns]