, , , ,
This study focuses on modeling and forecasting high-dimensional functional time series (HDFTS) with cross-sectional correlation and temporal dependence. A novel two-way functional median polish decomposition method is introduced to effectively decompose HDFTS into deterministic and time-varying components, providing robustness against outliers. The forecasting approach utilizes dynamic functional principal component analysis to generate accurate forecast curves for multiple populations by combining these forecasts with the deterministic components. The methodology is applied to age- and sex-specific mortality rates from the US, France, and Japan, showcasing its effectiveness in various sub-national regions. It outperforms several benchmark methods in terms of point and interval forecasts for multi-population mortality. However, the forecasting performance varies across countries due to their unique characteristics and population structures. Opportunities for further extension of this methodology include addressing the impact of outliers on principal component scores through robust functional principal component analysis or other robust time series methods. Additionally, additional levels of disaggregation such as cause-of-death or socioeconomic status could be integrated into the model with appropriate data availability, enhancing its accuracy and applicability in diverse mortality data scenarios.
Summary- The study looks at predicting patterns in lots of data over time that are related to each other and change with time.
- It introduces a new way to break down this data into different parts, making it stronger against unusual data points.
- The method uses a special analysis to make accurate predictions for different groups by combining forecasts with certain components.
- The approach was tested using information about how many people of different ages and genders die in the US, France, and Japan, showing it works well in different areas.
- It performs better than other methods in predicting death rates for different groups of people but does differently depending on the country.
Definitions
- Modeling: Studying patterns and relationships in data to make predictions.
- Forecasting: Predicting what will happen in the future based on current information.
- Functional time series: Data that changes over time and is related to each other.
- Correlation: How two sets of data are connected or related to each other.
- Temporal dependence: How one set of data depends on another set over time.
Introduction:
The study of mortality rates is crucial for understanding population dynamics and making informed policy decisions. However, traditional methods of modeling and forecasting mortality rates have limitations when dealing with high-dimensional data that exhibit cross-sectional correlation and temporal dependence. This research paper proposes a novel approach to address these challenges by introducing a two-way functional median polish decomposition method and utilizing dynamic functional principal component analysis for forecasting.
Methodology:
The first step in the proposed methodology is to decompose the high-dimensional functional time series (HDFTS) into deterministic and time-varying components using the two-way functional median polish decomposition method. This approach effectively deals with outliers, providing robustness to the model. The deterministic components capture long-term trends, while the time-varying components account for short-term fluctuations in the data.
Next, dynamic functional principal component analysis is applied to generate accurate forecast curves for multiple populations by combining these forecasts with the deterministic components. This allows for incorporating cross-sectional correlation and temporal dependence in the forecasting process, resulting in more accurate predictions compared to traditional methods.
Application:
To showcase its effectiveness, this methodology is applied to age- and sex-specific mortality rates from three countries - US, France, and Japan. These countries were chosen due to their diverse characteristics and population structures. The results demonstrate that this approach outperforms several benchmark methods in terms of point and interval forecasts for multi-population mortality.
Discussion:
While this methodology shows promising results, there are opportunities for further extension. One such opportunity is addressing the impact of outliers on principal component scores through robust functional principal component analysis or other robust time series methods. Additionally, integrating additional levels of disaggregation such as cause-of-death or socioeconomic status could enhance its accuracy and applicability in diverse mortality data scenarios.
Conclusion:
In conclusion, this research paper presents a novel approach for modeling and forecasting high-dimensional functional time series with cross-sectional correlation and temporal dependence. By utilizing two-way functional median polish decomposition and dynamic functional principal component analysis, this methodology provides robustness against outliers and accurately predicts mortality rates for multiple populations. Further research and application of this approach could greatly benefit the field of mortality forecasting.
We consider modeling and forecasting high-dimensional functional time series (HDFTS), which can be cross-sectionally correlated and temporally dependent. We present a novel two-way functional median polish decomposition, which is robust against outliers, to decompose HDFTS into deterministic and time-varying components. A functional time series forecasting method, based on dynamic functional principal component analysis, is implemented to produce forecasts for the time-varying components. By combining the forecasts of the time-varying components with the deterministic components, we obtain forecast curves for multiple populations. Illustrated by the age- and sex-specific mortality rates in the US, France, and Japan, which contain 51 states, 95 departments, and 47 prefectures, respectively, the proposed model delivers more accurate point and interval forecasts in forecasting multi-population mortality than several benchmark methods.
