This is a python project which using FedAvg to predict the future temperature.

environment configuration: Python 3.11 conda install pandas conda install numpy conda install scikit-learn conda install matplotlib pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121 #install torch in base by accidient pip install tqdm

parameter configuration： 客户端数量number of clients = 10 (so that each client has 300 data) 全局轮global epoch = 10 本地轮local epoch = 5 device = RTX 4060 mobile

net: batchsize = 64 Linear1 13 * 16 Linear2 16 * 1 activation function = sigmoid optimizer = SGD loss function = MSEloss reduction = 'mean' learning rate = 0.01