An important part of high-temperature plasma study is the determination of the electron temperature dynamics in the tokamak plasma. At spherical tokamaks, one can use Thomson scattering diagnostics as well as soft X-ray emission diagnostics (SXR). The capabilities of electron temperature measurement by the first diagnostics are limited by the repetition rate of laser pulses and their number in one tokamak discharge. Data of the second diagnostics are continuous in time and are determined by the time resolution of the detectors; however, obtaining the electron temperature using these data encounters a number of difficulties considered in this study. A method of combined processing of results of these diagnostics using machine learning algorithms was developed for overcoming these difficulties and applying the adVoprosy Atomnoi Nauki i Tekhniki, Seriya: Termoyadernyi Sintezages of both diagnostics. Training data include soft X-ray diagnostic data, hard X-ray diagnostic data, and CIII line emissivity diagnostic data. Thomson local scattering measurements were used as labels for supervised machine learning. The developed technique provides significant extension of the possibilities of determining the electron temperature at the Globus-M2 tokamak.