Ctrl+K

Model skill visualisation

import modelskill as ms

Model skill visualisation#

fn = 'data/SW/HKZN_local_2017_DutchCoast.dfsu'
mr = ms.model_result(fn, name='HKZN_local', item=0)

obs = [ms.PointObservation('data/SW/HKZA_Hm0.dfs0', item=0, x=3.9, y=52.7, name="HKZA"),
        ms.PointObservation('data/SW/HKZA_Hm0.dfs0', item=0, x=3.8, y=52.5, name="HKZA_2"),
        ms.PointObservation('data/SW/HKZA_Hm0.dfs0', item=0, x=3.5, y=52.6, name="HKZA_3"),
        ms.PointObservation('data/SW/HKNA_Hm0.dfs0', item=0, x=4.2420, y=52.6887, name="HKNA"),
        ms.PointObservation('data/SW/HKNA_Hm0.dfs0', item=0, x=4.2, y=52.6, name="HKNA_2"),
        ms.PointObservation('data/SW/HKNA_Hm0.dfs0', item=0, x=4.3, y=52.7, name="HKNA_3"),
        ms.PointObservation("data/SW/eur_Hm0.dfs0", item=0, x=3.2760, y=51.9990, name="EPL"),
        ms.PointObservation("data/SW/eur_Hm0.dfs0", item=0, x=3.2, y=51.9, name="EPL_2"),
        ms.PointObservation("data/SW/eur_Hm0.dfs0", item=0, x=3.3, y=51.95, name="EPL_3")
]

cc = ms.match(obs=obs, mod=mr)
cc

<ComparerCollection>
Comparers:
0: HKZA - Significant wave height [m]
1: HKZA_2 - Significant wave height [m]
2: HKZA_3 - Significant wave height [m]
3: HKNA - Significant wave height [m]
4: HKNA_2 - Significant wave height [m]
5: HKNA_3 - Significant wave height [m]
6: EPL - Significant wave height [m]
7: EPL_2 - Significant wave height [m]
8: EPL_3 - Significant wave height [m]

Data analysis#

cc.skill()
n bias rmse urmse mae cc si r2
observation
HKZA 397 0.097677 0.297203 0.280693 0.234103 0.965766 0.096848 0.924528
HKZA_2 397 -0.084598 0.294471 0.282057 0.224945 0.967457 0.097318 0.925909
HKZA_3 397 -0.055265 0.267680 0.261913 0.206410 0.971840 0.090368 0.938777
HKNA 386 -0.315380 0.447311 0.317210 0.341344 0.968323 0.102122 0.847042
HKNA_2 386 -0.263059 0.399579 0.300771 0.299503 0.970854 0.096830 0.877944
HKNA_3 386 -0.251149 0.393604 0.303065 0.290698 0.970244 0.097568 0.881567
EPL 66 -0.077263 0.229352 0.215947 0.194178 0.969118 0.083134 0.929378
EPL_2 66 -0.127012 0.251000 0.216493 0.217022 0.968078 0.083344 0.915417
EPL_3 66 -0.067380 0.227172 0.216950 0.190187 0.968860 0.083520 0.930714 
s = cc.skill()
type(s)

modelskill.skill.SkillTable

s.rmse.plot.bar();
_images/89999248b83592767a591f9303ffd8e3429d462172be21b6dfa4d501cdc5564b.png 
s['urmse'].plot.line();
_images/a36a167917c16363a0c2185d71026511a7caf424a74f4fa25d77eedd9071cc82.png

Custom plot#

All skill statistics are available in a dataframe, and in case you need a tailor-made plot, you can get data and use matplotlib to get exactly what you need.

import matplotlib.pyplot as plt

df = s.to_dataframe().sort_values('bias')
x = df.bias
y = df.index

plt.subplots(figsize=(6,6))
plt.scatter(x,y,marker='D',c='red',s=100)
plt.xlim(-0.35,0.35)
plt.axvline(0,linestyle='--')
plt.xlabel("Bias (m)")
plt.title(mr.name)
plt.grid()
_images/083fd12269f1dabf8bc806130c9ddd42596f503afda4cd853ad4e6dd422f37b3.png
Contents