如何优化此 Python 代码？

Question

I want to get the mean of the Year value of all IndicatorsCode of every country:

import numpy as np
import pandas as pd
datos = pd.read_csv("suramerica.csv")

media = list()
agricultura = list()
flag=0

paises = np.array(['Antigua and Barbuda','Argentina','Chile','Colombia'])
indicadores_agricultura = np.array(['EG.ELC.ACCS.RU.ZS','EG.NSF.ACCS.RU.ZS'])

for i in paises:
    for j in indicadores_agricultura:
        for k in range(len(datos)):
            if i==datos['CountryName'][k] and j==datos['IndicatorCode'][k]:
                flag=1
                media.append(datos['Year'][k])
    if flag==1:
        agricultura.append(np.array([i,np.mean(media)]))
        del media[:]
        flag=0
pd.DataFrame(agricultura,columns=['Paises','Agricultura y Desarrollo Rural'])

Here is a DataFrame of the result:

If you need access to the csv: Suramerica.csv

This code takes a long time to execute. Thanks for your time - any advice will be great.

Answer 1

There seems no need to traverse complete data for every combination. I am using a dict object to save required information. Then calculating np.mean using that. This will greatly enhance the execution speed. Here's code :

import numpy as np
import pandas as pd
datos = pd.read_csv("suramerica.csv")

agricultura = list()

output = {}


paises = np.array(['Antigua and Barbuda','Argentina','Chile','Colombia'])
indicadores_agricultura = np.array(['EG.ELC.ACCS.RU.ZS','EG.NSF.ACCS.RU.ZS'])


for k in range(len(datos)):
    cn = datos['CountryName'][k]
    indicator_code = datos['IndicatorCode'][k]
    # change1
    if cn not in output.keys():
            output[cn] = []
    if cn in paises and indicator_code in indicadores_agricultura:
        year = datos['Year'][k]

for o in output:
    # change2
    media = output.get(o)
    if not media:
        media = 0.0
    agricultura.append(np.array([o,np.mean(media)]))

output2 = pd.DataFrame(agricultura,columns=['Paises','Agricultura y Desarrollo Rural'])
print(output2)

Answer 2

I would start writing the loop this way:

for k, _ in enumerate(datos):
    cn = datos['CountryName'][k]
    ic = datos['IndicatorCode'][k]

    for i in paises:
        if i != cn:
            continue
        for j in indicadores_agricultura:
            if j == ic:
                flag = 1
                media.append(datos['Year'][k])

    if flag:
        agricultura.append(np.array([i,np.mean(media)]))
        del media[:]
        flag = 0