 In this study, we proposed and evaluated deep neural network, DNN, based load forecasting models for individual customer electricity consumption data. We found that DNNs outperformed other forecasting models such as a shallow neural network, SNN, the double-seasonal Holt Winters, DSHW, model, and an autoregressive integrated moving average, AREMA, model. Specifically, our DNN models achieved lower mean absolute percentage errors, MAPE, and relative root mean squared errors, RRMA-C, than the other models. This suggests that DNNs can be used to accurately predict electricity consumption at the individual level. This article was authored by Sung Hyun Ryul, Jaekun Oh, and Hongseek Kim.