使用Pandas数据框时用户定义的函数存在问题

Question

I have encountered an issue when calling a user defined function when using Pandas DataFrames. Stock market data is read in from a SQLite3 database in the following form:

date high low close

The following code sums high, low and close values for each row and adds a new column 'Sum' to df:

def Sum(h, l, c):
    return h+l+c

df.loc[:, 'Sum'] = Sum(df['high'], df['low'], df['close'])

             high    low  close     Sum
date
2018-01-23  80.65  78.25  79.45  238.35
2018-01-24  81.65  79.50  80.50  241.65
2018-01-25  81.70  80.25  81.10  243.05
2018-01-26  81.25  78.25  78.75  238.25
2018-01-29  70.95  62.25  64.15  197.35

However, if the function is changed to return the maximum value of high, low, close for each row in df an error ("ValueError: The truth value of a Series is ambiguous.") is generated.

def Max(h, l, c):
    return max(h, l, c)

df.loc[:, 'Max'] = Max(df['high'], df['low'], df['close'])

What is the issue with the Max function?

Answer 1

Jon Clements' comment is the way you should go. However, should you wish to perform more complex complications, pd.DataFrame.apply has this functionality:

Note I have renamed your function to avoid conflicts with built-in functions.

def max_df(h, l, c):
    return max(h, l, c)

df['Max'] = df.apply(lambda row: max_df(row['high'], row['low'], row['close']), axis=1)

This is a good post to learn about these options, if you cannot vectorise your calculation:

Difference between map, applymap and apply methods in Pandas