问题
I am very new to using R. I have searched regarding this but couldn't find any pointers.
I am doing cohort analysis and the data under Month1 is the number of new users that signed up for in that particular month. Month2 is the number of users that continued from previous month's sign up.
I have a set of data as follows
Month_Start_Date Month1 Month2 Month3 Month4 Month5 Month6
2010-01-01 10 12 11 9 3 15
2010-02-01 10 9 11 9 3 15
2010-03-01 10 9 7 9 3 15
2010-04-01 10 7 5 6 3 15
2010-05-01 10 8 6 3 4 15
I would like my data to be in the following format:
Month_Start_Date Month1 Month2 Month3 Month4 Month5 Month6
2010-01-01 10 0 0 0 0 0
2010-02-01 10 9 0 0 0 0
2010-03-01 10 9 7 0 0 0
2010-04-01 10 7 5 6 0 0
2010-05-01 10 8 6 3 4 0
My main data has data from 2008. I am filtering to take only from 2010. Hence, I would want the Month2, Month3, Month4, Month5 as 0 for 2010-01-01 and Month3, Month4, Month5 as 0 for 2010-02-01
回答1:
It seems like you want upper.tri to set the values to zero. Starting with
> df
# Month_Start_Date Month1 Month2 Month3 Month4 Month5 Month6
# 1 2010-01-01 10 12 11 9 3 15
# 2 2010-02-01 10 9 11 9 3 15
# 3 2010-03-01 10 9 7 9 3 15
# 4 2010-04-01 10 7 5 6 3 15
# 5 2010-05-01 10 8 6 3 4 15
we can replace the upper triangle with zeros with
> df[-1][upper.tri(df[-1])] <- 0
> df
# Month_Start_Date Month1 Month2 Month3 Month4 Month5 Month6
# 1 2010-01-01 10 0 0 0 0 0
# 2 2010-02-01 10 9 0 0 0 0
# 3 2010-03-01 10 9 7 0 0 0
# 4 2010-04-01 10 7 5 6 0 0
# 5 2010-05-01 10 8 6 3 4 0
来源：https://stackoverflow.com/questions/25857769/how-to-replace-column-values-for-cohort-analysis