A new model for learning-based forecasting procedure by combining k-means clustering and time series forecasting...

Area of Science:

• Data Science
• Machine Learning
• Time Series Analysis

Background:

• Traditional time series forecasting models often struggle with complex data patterns.
• Parameter tuning in forecasting algorithms can be a significant challenge, impacting accuracy.
• Clustering techniques offer potential for improving data segmentation and subsequent analysis.

Purpose of the Study:

• To propose a novel hybrid model for time series forecasting.
• To enhance forecasting accuracy by incorporating a clustering algorithm.
• To develop a learning-based procedure for automatic parameter estimation in forecasting.

Main Methods:

• A k-means clustering algorithm is utilized to group time series data.
• Forecasting is performed on the clustered data segments.
• A learning-based procedure is integrated to estimate forecasting parameters concurrently.
• The proposed model's performance is evaluated against existing forecasting algorithms.

Main Results:

• The proposed model demonstrated superior performance compared to other forecasting algorithms.
• Achieved the smallest mean squared error (MSE) of 13,007.91.
• Obtained an average improvement rate of 19.83% over baseline methods.

Conclusions:

• The integration of k-means clustering significantly improves time series forecasting accuracy.
• The learning-based parameter estimation enhances model robustness and reduces manual tuning.
• The proposed hybrid model offers a promising approach for complex time series forecasting tasks.