How to make my buffer operation in a python class faster/more efficient?

Question

I have a variable in a Class. self.temp = pd.DataFrame(columns = ['A', 'B'])

Not a coder by profession. Wrote a piece of code. Was running fairly fast. Made changes. Much slower now.

Added a variable in a class whose sole purpose was to be a buffer for operations on data from a csv. The following operations are carried out on it.

1. self.temp.loc[len(self.temp.index)] = [self.CALC.A.iloc[-1], self.CALC.B.iloc[-1]]
2. self.RESULT.loc[len(self.ATR.index)] = [self.temp.A.iloc[-1], self.temp.B.iloc[-1]]
3. self.temp = self.temp[0:0]

Is there a more efficient way to do this? Currently it is very time consuming (200,000+ rows to process).

Another line of code added was self.RESULT.drop(self.RESULT.tail(1).index,inplace=True)

Any and all ways to make the code more efficient after these changes will be much appreciated. :)