Aiming to reduce the impact of outliers in calculating mean of each row, I want to develop a ruled based mean function. In the below df for example:
2 3 4 5 3 4 5 6 80 3 3
4 5 2 4 5 5 2 4 4 56 3
2 33 4 44 3 4 5 6 80 3 2
1 3 4 1 3 4 5 6 80 3 3
66 66 2 4 5 66 2 4 3 56 3
2 33 4 44 3 4 5 6 80 3 2
22 55 33 33 44 25 34 38 31 34 2
So for each row, in a new dataframe (the current one should remain untouched), I want to replace all the numbers that are greater with {current mean(row)} and then get a new mean(row) with the replaced numbers.
The new mean should be inserted in a new column to be compared with the current mean. Any new scientific method to address the same problem is highly appreciated.