As observed, the climatic variables have a poor influence on energy consumption. Moreover, the occupancy data is not readily available as the cluster of these three buildings has many openings for occupants to enter and exit. All such factors make it difficult to measure an exact occupancy count. For these reasons, the focus is given to investigate on the internal energy consumption pattern. This means that rather than developing a model based on external factors
as inputs (like climatic variables, occupancy etc.), the internal consumption pattern is assessed and the daily energy consumption for cooling is taken as inputs to predict the forthcoming days. The scale of data analyzed is on the yearly level. For this, average daily consumption is taken as an assessment element and the daily energy consumption is taken as inputs. Moreover, the weekend energy data is excluded from this analysis as the building is officially closed during
these times. Excluding the weekends, there are about 250 data points (one for each day) and each value corresponds to the average energy consumption value for that day. The first level of analysis entails predicting energy consumption data for next successive day using the past data for previous three days. A graphic example of this is presented in Fig.4.