Abstract:This study proposes a prediction model considering external weather and holiday factors to address the issue of accurately predicting urban taxi travel demand caused by complex data and numerous influencing factors. The model integrates the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Convolutional Long Short Term Memory Neural Network (ConvLSTM) to predict short-term taxi travel demand. The CEEMDAN decomposition method effectively decomposes time series data into a set of modal components, capturing sequence characteristics at different time scales and frequencies. Based on the sample entropy value of components, secondary processing of more complex sequence components after decomposition is employed to reduce the cumulative prediction error of component sequences and improve prediction efficiency. On this basis, considering the correlation between the spatiotemporal trends of short-term taxi traffic, a ConvLSTM neural network model with Long Short Term Memory (LSTM) time series processing ability and Convolutional Neural Networks (CNN) spatial feature processing ability is constructed to predict the travel demand for urban taxis. The combined prediction model is tested on a taxi travel demand dataset in a certain area of Beijing. The results show that the CEEMDAN-ConvLSTM prediction model outperforms the LSTM, Autoregressive Integrated Moving Average model (ARIMA), CNN, and ConvLSTM benchmark models in terms of Symmetric Mean Absolute Percentage Error (SMAPE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R2 metrics. Notably, the SMAPE metric exhibits a remarkable decline of 21.03% with the utilization of our proposed model. These results confirm that our study provides a highly accurate and valid model for taxi travel demand forecasting.

Multivariate Time-Series Forecasting Model: Predictability Analysis and Empirical Study

Experimental Study of Multivariate Time Series Forecasting Models

An Entropy-Based Approach for Evaluating Travel Time Predictability Based on Vehicle Trajectory Data

Model-free quantification of time-series predictability

Forecasting Traffic Time Series with Multivariate Predicting Method.

An Information Theory Based Method for Quantifying the Predictability of Human Mobility

Parallel Extraction of Long-term Trends and Short-term Fluctuation Framework for Multivariate Time Series Forecasting

Human Mobility: Prediction and Predictability

Measuring the Gap Between the Maximum Predictability and Prediction Accuracy of Human Mobility

Predictability in Human Mobility: from Individual to Collective (vision Paper)

A Pattern Discovery Approach to Multivariate Time Series Forecasting

On Predictability of Time Series

A Neural Networks Based Method for Multivariate Time-Series Forecasting

Enhanced Time Series Predictability with Well-Defined Structures

An Attentive Copula-based Spatio-Temporal Graph Model for Multivariate Time-Series Forecasting

Beyond the Limits of Predictability in Human Mobility Prediction: Context-Transition Predictability

Predicting Taxi and Uber Demand in Cities: Approaching the Limit of Predictability

A Combination Prediction Model for Short Term Travel Demand of Urban Taxi

Exploring Progress in Multivariate Time Series Forecasting: Comprehensive Benchmarking and Heterogeneity Analysis

Diff-MGR: Dynamic Causal Graph Attention and Pattern Reproduction Guided Diffusion Model for Multivariate Time Series Probabilistic Forecasting

Multivariate Temporal Convolutional Network: A Deep Neural Networks Approach for Multivariate Time Series Forecasting