1. Science
  2. Publications
  3. Information Processing Systems
  4. 6(87)'2010
  5. CLUSTERIZATION OF DATA FOR FORECASTING FOR LANDSLIDE PROCESSES OF THE SOUTHERN COAST OF CRIMEA

CLUSTERIZATION OF DATA FOR FORECASTING FOR LANDSLIDE PROCESSES OF THE SOUTHERN COAST OF CRIMEA

V.N. Taran
Annotations languages:

It is suggested at the analysis of landslide processes of the Southern Coast of Crimea to examine the results of supervisions in a standard kind on purpose to bring them over to one dimension, by means of gravimetric coefficients to unite all entrance factors in one variable: middle weighted. There is offered ROMW-model (regression onefactor middle weighted). It is set that at the ascending ordering of independent factor a resulting index is also put in order. The ranked data are broken up on three large clusters, limits for an entrance factor, at which a resulting index acquires one of three values, are set: little amount of landslides, AV and large. Reasons of origin of divergences are certain between prognosis and real data for the model of ROMW is the level of the inlaid facilities in strengthening of slopes. For other models the presence of this conformity to law is not observed. By means of criteria of estimation of models and estimation of prognosis five models, offered before, are analyzed, and model of ROMW, drawn conclusion, that all models are suitable for forecasting. Figs: 5. Tabl.: 4. Refs: 12 titles.
Keywords: forecasting of landslide processes of the Southern Coast of Crimea, regressive analysis, model estimation and prognosis estimation, clusterization