Electricity Consumption Prediction based on Geo- Demographic Factors

The task of predicting electricity usage is crucial for energy management and planning. The intricate connections and patterns of power consumption statistics, which are influenced by a variety of geodemographic parameters like weather, population, income, etc., are difficult to record. In this study, we provide a novel model that combines Long Short-Term Memory (LSTM) and an attention mechanism to forecast electricity consumption based on geodemographic parameters. The relevance and significance of the incoming sequence data are weighted differently by the attention mechanism. The LSTM network generates the projected values after learning the temporal and spatial correlations of the data on electricity use. On two real-world datasets with various spatial granularities—appliance-level and household-level—we test our model. The outcomes demonstrate that our approach outperforms cutting-edge techniques.