如何将 function 应用于 pandas dataframe 中一列的每一行？

Question

I have a dataframe df of stock prices of length ~600k, which I downloaded from here .

I have renamed the last column name from 'Name' to 'ticks', and created a new blank column called 'Name':

df = df.rename(columns={'Name': 'Ticker'})
df['Name'] = ''

I have written the following function to return the company name for a given ticker symbol:

! pip3 install yfinance
import yfinance as yf

def return_company_name(ticker):
    return yf.Ticker(ticker).info['longName']

return_company_name('MSFT')
>>> 'Microsoft Corporation'

Now, I want to populate the column 'Name' with the company name of the corresponding ticker symbols. For that, I have written the following lambda function:

df.Name = df.Ticker.apply(lambda x: return_company_name(x))

But this last line of code just keeps on running. Is there something going wrong? If yes, how do I fix it?

I tried the same with map instead of apply , but same result.

Answer 1

First, you don't need a lambda or apply .

 df.Name = df.Ticker.map(return_company_name)

Is better. Second, as pointed out by others, this is grotesquely inefficient. You are making the call 600000 times, even though your number of tickers is much smaller. The following sledgehammer approach will work:

class my_return():
     def __init__(self):
         self.tickdict = {}
     def __call__(self, ticker):
         ans = self.tickdict.get(ticker, None)
         if ans is not None:
             return ans
         else:
            self.tickdict[ticker] = return_company_name(ticker)
            return self.tickdict[ticker]

Then map my_return on your ticker column.

Answer 2

Looking at the source from yfinance you can see here that the get_info method calls _get_fundamentals which in turn seems to do quite a few API calls to different sites to get the information it needs.

Since this is executed for every row you run into some trouble as the sites might rate limit you. Maybe you could do a prestep of getting all the unique names and then looking them up once and saving them in some kind of lookup CSV or the like

Answer 3

You can use pandas.apply() to apply a function to each row/column in Dataframe.

You also can use lambda function to each column. For example:
modDfObj = dfObj.apply(lambda x: x + 10)

Another example (Here, it only applies the function to the column z ):

modDfObj = dfObj.apply(lambda x: np.square(x) if x.name == 'z' else x)