Analysis of the relationship between LSTM network traffic flow prediction performance and statistical characteristics of standard and nonstandard data

Abstract

The effectiveness of road traffic control systems can be increased with the help of a model that can accurately predict short-term traffic flow. Therefore, the performance of the preferred approach to develop a prediction model should be evaluated with data sets with different statistical characteristics. Thus a correlation can be established between the statistical properties of the data set and the model performance. The determination of this relationship will assist experts in choosing the appropriate approach to develop a high-performance short-term traffic flow forecasting model. The main purpose of this study is to reveal the relationship between the long short-term memory network (LSTM) approach's short-term traffic flow prediction performance and the statistical properties of the data set used to develop the LSTM model. In order to reveal these relationships, two different traffic prediction models with LSTM and nonlinear autoregressive (NAR) approaches were created using different data sets, and statistical analyses were performed. In addition, these analyses were repeated for nonstandardized traffic data indicating unusual fluctuations in traffic flow. As a result of the analyses, LSTM and NAR model performances were found to be highly correlated with the kurtosis and skewness changes of the data sets used to train and test these models. On the other hand, it was found that the difference of mean and skewness values of training and test sets had a significant effect on model performance in the prediction of nonstandard traffic flow samples.

RF‐LSTM‐based method for prediction and diagnosis of fouling in heat exchanger

“…Based on feedback, Neural network models are widely categorized as the feed‐forward neural network (FFNN) and recurrent neural network (RNN). FFNN handles the spatial domain data without considering the temporal information 19 . RNN has a loopback network architecture that considers both the time and sequential interdependencies among the data 44 .…”

“…It can solve vanishing and exploding gradient problems, which usually occur while handling longer sequences. It also minimizes the propagation of local errors 19 . Local memory in the LSTM network enables to update the recent information and forget the unwanted historical data 46 …”

“…Hydraulic aspects can take up both the fouling built‐up and variability in the flow rate. Hence, the thermal and hydraulic aspects are the reliable indicators of fouling in the HE 19 …”

A hybrid deep learning approach by integrating extreme gradient boosting‐long short‐term memory with generalized autoregressive conditional heteroscedasticity family models for natural gas load volatility prediction

“…To measure the accuracy of the prediction model, this study uses mean square error (MSE) and root mean square error (RMSE), 43 and the index equations are shown in Equations ( 32)- (33). These two evaluation indicators are used because it is considered that in the neural network, the MSE can effectively converge even if a fixed learning rate is used, and the gradient of the MAE update is always the same, which is not conducive to the learning of the model.…”

A novel network traffic combination prediction model

“…The development of deep learning brings a new direction to network traffic prediction 25 ; common prediction methods include recurrent neural network, long short‐term memory (LSTM), 26 gated recurrent unit (GRU), 27 convolutional neural network (CNN), 28 deep belief networks (DBN), 29 and extreme learning machine (ELM) 30 . Deep learning is very suitable for network traffic prediction, but it also has some defects, for example, the structure is not uniform, the hyperparameters are difficult to determine and need the optimization algorithm, and the computing cost is increased with the help of GPU computing power.…”