0
$\begingroup$

I have 10k rows of multi-classification (x1..x27,y), size of dataframe is: 28*10k and its single output feature is imbalanced meaning that it is zero in 98.5% and the other 1.5% * 10k = 150 values are listed in the following

0.68904
0.99
0.99
1
1.38504
1.38504
1.38504
1.38504
1.38504
1.38504
1.38504
1.38504
1.38504
1.38504
1.99
1.99
1.99
1.99
1.99
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2.08104
2.08104
2.08104
2.49
2.49
2.49
2.533026481
2.664213656
2.734707229
2.77008
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.77704
2.99
2.99
2.99
2.99
2.99
2.99
3
3.47304
3.47304
3.749138211
3.98
3.99
3.99
4
4
4.16208
4.16208
4.16208
4.16208
4.48
4.48
4.85808
4.97
4.99
4.99
4.99
4.99
4.99
4.99
5.97
5.97
6
6
6
6
6.24312
6.48
6.95304
6.97
6.97
6.986
7
7.38504
7.62816
7.63512
7.98
7.98
8
9
9.03408
9.73008
9.96
9.99
9.99
11
11.30527972
11.81112
11.98
12
12
12.96
13.89912
13.94
13.96
14
15
15
17.94608
17.97
18
19.02608
19.97
20.10048
20.93
22.2024
23.70007731
24.98
25
30
36
37.95
46.88
48.93
56.37
64
65.94
67
68.43
77.84
107.91
116.81
206.81
429.94
2000

as it is shown its range is different from 0 to 2000! How we can binarized the output class(y)? I uploaded a sample rows of my data here

I implemented RF,XGboost but their accuracy is 98.8% which is not high! How I can improve my results? Which parameters and what are the range of parameters are your suggestion to tune better? I used default values but somehow playing with some parameters, it did not changed!

$\endgroup$
  • $\begingroup$ It seems like a regression output, not a class or labels, right? $\endgroup$ – Grzegorz Oct 20 at 12:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.