We predict the indoor temperature in a building by using the outdoor tempera- ture through one temperature sensor, which is located outside the building, an initial indoor temperature, and a normal distribution of students arrival to the building using their class schedule. We also have an indoor temperature sen- sor, which is used to collect the ground truth. Our model predicts the indoor temperature based on when the entrance door to the building is opened and for how long. To predict the temperature, the model estimates the amount of air that escaped (based on the temperature difference between the indoor and outdoor and the time the door was opened). The estimation of the air escaped is used to compute the indoor temperature after the air escaped. The idea is to have a normal distribution for each class that predicts the number of students arriving to class, sum the number of students and coming to the building one after the other to get the number of seconds the door is opened every minute and predict the indoor temperature using our model. We use linear regression to increase the accuracy. We also use linear and quadratic b-splines to predict the indoor temperature. We compare all these methods to the ground truth.