Research Article

Characterization of Meteorological Drought Using Monte Carlo Feature Selection and Steady-State Probabilities

Table 7

Relative importance (RI) values computed for three selected drought indices. The varying RI values can be observed for various indices. For example, for SPI, in the Astore Station, the RI value is 0.1385, and in Bunji, the RI value is 0.1323. The RI values of 0.1015, 0.1212, 0.1314, and 0.1295 are computed for Gupis, Chilas, Gilgit, and Skardu, respectively. The RI value in Astore is higher than other stations for SPI. In SPEI, RI value is 0.2200 for Astore Station. The RI value of 0.5524 is calculated for Bunji Station. For Gupis, Chilas, Gilgit, and Skardu, the RI values are 0.3547, 0.5525, 0.7617, and 0.5548, respectively. Based on SPEI, the Gilgit receives higher weights. In SPTI, the Astore has a higher RI value of 0.1922.

Relative importance (RI)
SPISPEISPTI

Astore0.13850.22000.1922
Bunji0.13230.55240.1322
Gupis0.10150.35470.0963
Chilas0.12120.55250.1298
Gilgit0.13140.76170.1306
Skardu0.12950.55480.1547