pandas 列使用 function 计算，包括字典查找，“系列”对象是可变的，因此它们不能被散列

Question

I am aware there are tons of questions similar to mine, but I could not find the solution to my question in the last 30 Minutes of looking through dozens of threads.

I have a dataframe with hundereds of columns and rows, and use most columns within a function to return a value thats supposed to be added to an additional column.

The problem can be broken down to the following.

lookup = {"foo": 1, "bar": 0}

def lookuptable(input_string, input_factor):
    return lookup[input_string] * input_factor

mydata = pd.DataFrame([["foo", 4], ["bar",3]], columns = ["string","faktor"])
mydata["looked up value"] = lookuptable(mydata["string"], mydata["faktor"])

But this returns:

TypeError: 'Series' objects are mutable, thus they cannot be hashed

Is there a way to avoid this problem without, restructuring the function itself?

Thanks in advance!

Answer 1

Try this:

lookup = {"foo": 1, "bar": 0}

def lookuptable(data):
    return lookup[data["string"]] * data["faktor"]

mydata = pd.DataFrame([["foo", 4], ["bar",3]], columns = ["string","faktor"])
mydata["looked up value"] = mydata.apply(lookuptable, axis=1)

print(mydata)

    string  faktor  looked up value
0   foo        4        4
1   bar        3        0

Answer 2

Besides of using .apply() , you can use list comprehension with .iterrows()

mydata["looked up value"] = [lookuptable(row[1]["string"], row[1]["faktor"]) for row in mydata.iterrows()]

Answer 3

Your functions accepts 2 parameters, a string and a integer. But you provide 2 pandas series to the function instead. You can iterate through the dataframe however and provide the function with the parameters (row-wise) by using .apply() .

mydata["looked up value"] = mydata\
.apply(lambda row: lookuptable(row["string"], row["faktor"]), axis=1)

Answer 4

You can do this without function -

import pandas as pd
lookup = {"foo": 1, "bar": 0}
mydata = pd.DataFrame([["foo", 4], ["bar",3]], columns = ["string","factor"])
mydata["looked up value"] = mydata['string'].map(lookup) * mydata['factor']