连接每行值直到 python dataframe 中的 NaN 值

Question

I am very new to python and I am trying to find a way to solve the below issue, if anyone knows a solution to this query please do help.

Thanks in advance!

I want to concatenate each row values until a nan value and then store it as the respective row value of a new column.

Below is an example to support my query: The input dataframe is as below:

df = pd.DataFrame({"student_name":['mike','maria','alex','mary','shirin'],"student_id":[1,2,3,4,5], "a1":[70,np.nan,64,78,79],"a2":[65,75,72,np.nan,61],"a3":[82,79,80,99,20],"a4":[90,34,56,89,67],"a5":[78,89,90,90,55],"a6":[55,78,88,77,84]})

I have attached a picture of the input dataframe below:

My need is:

1. I want to concatenate each corresponding row values of columns a1...an and then store the value to tag column. 2)If the row contains a nan or an empty value then concatenating should stop at that value.

For the above input dataframe there are a1 to a6 columns, hence the desired output is as shown in the below picture.

Answer 1

Here's one quick Python 3 solution for your DataFrame (assuming your table will be non-empty as a whole for using this code snippet)

# ... following your code up to df= padas.DataFrame() line 

tags_ = []
for rowIndex in range(len(df[df.columns[0]])):
    tag_ = ""
    for col in df:
        if col.startswith('a'):
            try: tag_ += str(int(df[col][rowIndex]))
            except: break
    tags_.append(tag_)

df.insert(len(df.columns), "tag_", tags_)

Here, tags_ is just a list to store your string concatenated values per row until a numpy.nan is encountered (though code doesn't check for it specifically) The nested loops iterate through every row and column of your DataFrame, and non-empty DataFrame columns is required to assure the identification of rows in it, at the initial rowIndex for-loop.

df.insert(<location>, <column_name>, <values>) finally inserts the desired tag_ column as the ending column of your DataFrame.

Hopefully, it helps. Any corrections against it are most welcome.