The book presents practical guidelines to dependencies reconstruction on small samples. Descriptions of theoretical basis and computational algorithms are followed by program packages in FORTRAN-IV for pattern recognition, multiple regression and ill-posed inverse problems solution. All algorithms and programs implement the unified principle of structural minimisation of loss function. This principle allows to find optimal relationship between complexity of solution and amount of empirical data.

The book is oriented to professionals in the flied of statistics implementation in economy, medicine, geology, technical systems and scientific investigations.