When studying variations of cosmic ray intensity, by the use of muon telescopes located deep in the atmosphere it is necessary to take into account changes in atmospheric parameters, mainly pressure and temperature. The density distribution of temperature coefficients of the atmosphere muon intensity needs to be estimated from observations. To this purpose, the method of principal components regression and meth-ods of projection to latent structures (PLS-1 and PLS-2). We used data of continuous recording of muons, as well as Novosibirsk 2004–2010 aerological data. As shown by comparing results, PLS-2 method allows us to esti-mate the density distribution of muon intensity temperature coefficients with minimal errors.