Energy loss represents a traditional key objective in the optimal operation and planning of electrical networks, and various estimation methods have been studied. In this study, two formulas are proposed for calculation of the loss factor (LSF) to improve the classical LSF method based on the minimum load factor (MLF) and the LF. The former is an approximate formula, whose accuracy is good enough for engineering calculations. While the latter is an empirical quadratic equation determined by the statistical analysis, which is more accurate to estimate losses. To conduct a complete feasibility study for project practices, a large amount of measurement data is used to calculate energy losses in a district of Guangdong using the classical LSF method and the improved LSF (ILSF) method. Results of statistical analyses indicate that the real data fall in the proposed three-dimensional region and the use of MLF can help improve accuracy in the energy losses estimation. The classical and ILSF methods are used to estimate the effect in loss reduction by inserting a distributed generation in a 43-bus distribution network, and the candidate bus can be identified effectively.