CODEWITHSUNDEEP
pythonpandas
johnrogeksu
johnrogeksu

Reputation: 85

Adding a new column to my dataframe inserts an NaN value instead of desired value

When I try to add three new columns they all initialize with NaN values instead of the desired values. However if I perform some logic after the empty dataframe is initialized then the values do not get initialized as NaN values.

This does not work


    auction = pandas.DataFrame()

    //These values initialize to NaN 
    auction['Date'] = dataFrame['report_date'].loc[0]
    auction['Market Location'] = dataFrame['market_location_name'].loc[0]
    auction['State'] = dataFrame['market_location_state'].loc[0]

This works


    auction = pandas.DataFrame()

    for i in range(4):
        //logic to initialize a few columns in auction

    //These values initialize to the desired values
    auction['Date'] = dataFrame['report_date'].loc[0]
    auction['Market Location'] = dataFrame['market_location_name'].loc[0]
    auction['State'] = dataFrame['market_location_state'].loc[0]

Its not a huge deal becuase I can always use insert to get the columns in the desired order but I am confused on why it isnt working with the first block of code.

Upvotes: 1

Views: 1338

Answers (1)

Ben.T
Ben.T

Reputation: 29635

I guess it is mostly that in the case that you called "does not work", the problem is that the dataframe has no index and initializing a column with a single value just create the column name. While in the "case that works", you already have index in your dataframe before doing auction['Date'] = dataFrame[... and in this case, the value is assign to each element in the column. with a simple example:

# example
dataFrame = pd.DataFrame({'a':list('mn'),
                          'b':range(2)})
print (dataFrame)
   a  b
0  m  0
1  n  1

Now case that "works"

auction = pd.DataFrame()
#initialize first with an iterable: like a list or a column from another dataframe
auction['c'] = range(2) #or auction['c'] = dataFrame['b']
auction['d'] = dataFrame['a'].loc[0]
print (auction)
   c  d
0  0  m
1  1  m

but the case that "does not work":

auction = pd.DataFrame()

auction['d'] = dataFrame['a'].loc[0]
print (auction) #the result is a empty dataframe with column d
# Empty DataFrame
# Columns: [d]
# Index: []

#now add a column from an iterable
auction['c'] = range(2) #or auction['c'] = dataFrame['b']
print (auction) #the column d did not have any value so filled with nan
#      d  c
# 0  NaN  0
# 1  NaN  1

Upvotes: 1

Related Questions