Feature Importance of XGBoost

The feature importance analysis for the XGBoost model highlights the relative contribution of each feature to model predictions. Below table lists all 88 variables along with their importance, cumulative importance, and rank for XGBoost in RECLAIM.

Feature importance of all 88 variables from XGBoost.

Rank	Variable	Feature Importance (%)	Cumulative Importance (%)
1	log_OBC	25.580140	25.580140
2	log_LCWB	18.123307	43.703447
3	log_CA	12.262684	55.966131
4	log_LCM	5.976027	61.942159
5	log_LCG	5.152744	67.094903
6	log_LCSG	4.975100	72.070003
7	log_LCC	1.642522	73.712525
8	log_LCS	1.506769	75.219294
9	log_SA_std	1.341006	76.560301
10	log_O_std	1.127022	77.687323
11	log_LCT	1.101582	78.788904
12	log_SA_mean	1.008979	79.797883
13	log_LCHV	0.887762	80.685645
14	log_LCAS	0.792153	81.477799
15	log_LCBS	0.718496	82.196295
16	log_RP	0.714475	82.910770
17	log_rain_per_area	0.667992	83.578762
18	R_SA_cap	0.602455	84.181217
19	SLOP	0.579330	84.760547
20	NSSC1_cv	0.557334	85.317881
21	SA_skew	0.520467	85.838348
22	log_DCA	0.514445	86.352793
23	MRB	0.478671	86.831465
24	log_MAI	0.433758	87.265223
25	log_FL	0.400659	87.665882
26	DLC	0.385051	88.050933
27	NSSC1_std	0.383057	88.433989
28	log_RA	0.350518	88.784507
29	R_shrub_bare	0.332395	89.116902
30	log_SOUT	0.318822	89.435724
31	R_coarse_sand	0.303878	89.739602
32	AGE	0.302968	90.042569
33	log_MAO	0.301705	90.344274
34	wind_std	0.290478	90.634752
35	LON	0.289763	90.924515
36	log_SA_mean_clip	0.288825	91.213340
37	ASP	0.286191	91.499531
38	SA_cv	0.283685	91.783217
39	log_PAI	0.262242	92.045458
40	E_std	0.259516	92.304975
41	I_cv	0.231728	92.536703
42	wind_cv	0.230122	92.766825
43	COAR	0.213779	92.980605
44	#_rain_above_100	0.211475	93.192080
45	#_rain_above_10	0.211080	93.403159
46	NSSC2_max_persis	0.210585	93.613745
47	log_ESR	0.202541	93.816285
48	NSSC1_skew	0.201005	94.017290
49	ELEV	0.200103	94.217393
50	NVGF	0.199504	94.416897
51	NSSC1_kurt	0.196393	94.613291
52	log_LCSV	0.194322	94.807613
53	SA_above_90	0.183946	94.991559
54	NSSC2_mean	0.183660	95.175219
55	#_rain_above_50	0.180766	95.355984
56	wind_skew	0.176269	95.532254
57	AECC	0.174868	95.707121
58	rel_SA_mean_clip	0.172370	95.879491
59	wind_mean	0.172309	96.051800
60	wind_kurt	0.168957	96.220757
61	O_cv	0.168843	96.389600
62	R_treaa_bare	0.168720	96.558320
63	CURV	0.165235	96.723555
64	SAND	0.158673	96.882228
65	SILT	0.156815	97.039043
66	NSSC1_mean	0.156760	97.195803
67	log_ROBC	0.153572	97.349375
68	I_max_persis	0.152358	97.501733
69	I_above_90	0.151440	97.653172
70	AECS	0.148747	97.801919
71	log_MAR	0.146104	97.948023
72	HILL	0.144328	98.092351
73	CLAY	0.143629	98.235980
74	log_HGT	0.141741	98.377721
75	log_SIN	0.137931	98.515653
76	log_ECLR	0.137038	98.652691
77	tmin_mean	0.126585	98.779275
78	log_TE	0.125490	98.904765
79	log_I_std	0.123198	99.027963
80	BULK	0.121505	99.149468
81	NSSC2_above_90	0.120844	99.270313
82	tmax_mean	0.114328	99.384641
83	log_GC	0.113850	99.498491
84	log_SA_kurt	0.105192	99.603683
85	LAT	0.103628	99.707311
86	log_RT	0.100726	99.808037
87	E_mean	0.098133	99.906170
88	AECI	0.093830	100.000000