CasiaFan / Time_seires_prediction_using_lstm

trainingset_selection.py is used to format input data to training data set. It will return files of those enterprises satisfying the screening rule.

neural_network_run.py is used to train neural network for time series prediction. It returns a lstm model and evaluation results of the model.

preprocessing.py provided some functions required in neural_network_run.py, training_set_selection.py and dataset_merging.py.

optional script

dataset_merging.py is used to merge similar enterprises data set (or enterprises whose data is within similar range) based on Q-Q plot and ks-statistic. It will return merged cluster data set files and summary file to illustrate enterprises in each cluster.