The Smart Grid is a key development of modern power grids that includes among other features, greater communication between power consumers and producers. While the investment required to implement Smart Grid technology is substantial, some cities have begun to implement it. Specifically, individual devices themselves can now communicate, allowing researchers to access device level data. Traditional infrastructure has only allowed data to be collected and analyzed at the apartment and home level. As some devices use far more power than others, this allows researchers to work at a finer unit level than previous work. The Pecan Street Dataset is made up of four cities that use a smart grid, in Texas, California, and New York with data collected on over a thousand homes. With temperature data from the nearest airport, we were able to look at the effect on time and temperature on not only the total power consumption of a domicile, but the individual devices such as air conditioners, refrigerators, etc. Homes with solar panels installed also had their net solar energy production included. Understanding how demand for power spikes is very important for the cost of power and the stability of a power grid. Despite the prevalence of renewable energy sources in many areas, they cannot often be solely relied upon in many climates. Thus, natural gas and coal power plants remain key for power grids to meet demand. One of the major issues with managing a large power grid is predicting the peak load. Battery technology is not effective at storing large amounts of electricity, requiring power plants to produce on an as needed basis. If there’s not enough power to go around, subsections of the grid must be disconnected. Often, companies that distribute power must pay severe penalties if they require more energy from the providers than they expected. Thus, predicting the peak load is an important issue for the stability of the power grid, as well as to power companies themselves. Previous literature sponsored by the Environmental Protection Agency has investigated the peak load of various devices using cost/benefit analysis. Our work goes further by using a Convolutional Neural Network (CNN) to investigate when peaks occur for specific types of devices, and the effect of temperature. Preliminary results indicate the temperature is a key predictor for air conditioners, HVAC devices, cars.