Table B.1
Summary table of the variables
variable | Mirror | Flip | Reco | γ/h-Sep. | A-stat. |
---|---|---|---|---|---|
Hillas-image-amplitude | x | Yes | x | 2.422 | |
Hillas-local-distance | 0.014 | Yes | 0.033 | x | |
Hillas-npix | 0.018 | Yes | 0.058 | 0.427 | |
Hillas-length | x | Yes | 0.045 | 6.637 | |
Hillas-width | x | Yes | 0.241 | 58.172 | |
Hillas-skewness | ×(−1) | 0.263 | Yes | 0.013 | 8.476 |
Hillas-kurtosis | x | Yes | 0.008 | 20.770 | |
Hillas-length-over-logsize | 0.008 | Yes | 0.021 | 5.819 | |
Hillas-logdensity | x | No | 0.066 | 31.479 | |
xhead | −xtail | x | No | 0.011 | 6.259 |
xcenter | x | No | x | x | |
xtail | xhead | x | No | 0.004 | 4.478 |
yhead | ytail | x | No | x | x |
ycenter | x | No | x | x | |
ytail | yhead | x | No | x | x |
widthhead | widthtail | 0.028 | No | 0.046 | 32.785 |
widthcenter | x | No | 0.015 | 20.559 | |
widthtail | widthhead | 0.026 | No | 0.008 | 29.886 |
charge-asymmetry | ×(−1) | x | Yes | x | x |
kink-major | x | No | 0.007 | 14.610 | |
kink-major-GOF | x | No | 0.026 | 34.703 | |
kink-minor | ×(−1) | 0.482 | No | 0.011 | 17.520 |
kink-minor-GOF | 0.012 | No | 0.016 | 7.115 | |
roughness | x | No | 0.006 | 13.753 | |
spatial-auto-correlation | x | No | 0.012 | 7.860 | |
profile-gradient | x | No | 0.186 | 11.150 | |
profile-gradienthead | profile-gradienttail | 0.013 | No | 0.005 | 9.967 |
profile-gradienttail | profile-gradienthead | 0.013 | No | 0.004 | 5.542 |
profile-gradient-GOF | x | No | 0.007 | 39.337 | |
std-over-mean | x | Yes | 0.057 | 9.958 | |
time-gradient-major | ×(−1) | x | Yes | x | 809.564 |
time-gradient-major-GOF | x | No | x | 24190.345 | |
time-gradient-minor | x | No | x | 2303.658 | |
time-gradient-minor-GOF | x | No | x | 19008.329 | |
fraction-top-1+2 | x | No | 0.011 | 0.519 | |
fraction-top-1 | x | No | 0.009 | 1.086 | |
npix>10 | x | No | 0.018 | 0.574 | |
npix>30 | x | No | 0.042 | 5.133 | |
distance-top-1 | x | No | 0.006 | 1.007 | |
distance-top-2 | x | No | 0.003 | -0.408 | |
distance-top-1-2 | x | No | 0.006 | 2.619 |
Note. The “Mirror” column shows how the variables behaves when being flipped, i.e., no change, change of sign, or change of variable (with sign change). Numbers in the columns for the flip and gamma-hadron separation training indicate the relative variable importance, whereas “x” denotes that the variable is not used for the respective training. For the direction and energy reconstruction the choice of variables is identical and no importance is estimated from the neural networks. The last column “A-stat.” gives the Anderson-Darling test statistics from Sect. 2.1.5.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.