The problem of indoor temperature prediction for further application in model predictive control synthesis was considered. The introduction of time delay for weather conditions was used to develop a simplified model of the temperature's dynamics that simulates thermal inertness of the building shell. The procedure of numerical estimation of the time delay was proposed on the basis of correlation analysis. Dataset of continuous measurements of indoor temperature and weather conditions was used in the experiment. The data were collected by the wireless network of smart sensors and local meteostation. Splitting the data into training and test samples was used to validate the accuracy of the prediction.