A Retrospective Forecast Audit of Recorded LSTM and ARIMA Inflation Projection Paths for Vietnam, October 2024 to September 2025

This study audits the recorded twelve-month LSTM and ARIMA paths for year-over-year consumer price inflation in Vietnam over October 2024–September 2025. Fixed projections are compared with official outcomes and with persistence and seasonal-naive benchmarks reconstructed from observations available after the release on 6 October 2024. LSTM has a path RMSE of 5.524 percentage points, compared with 1.217 for ARIMA, 0.863 for seasonal naive, and 0.596 for persistence. The January 2025 LSTM projection of 20.66% contrasts with the realized inflation of 3.63% and contributes 79.22% of its total squared error. Removing that month in a secondary influence check leaves the descriptive ranking unchanged. Original LSTM and ARIMA issue dates and future-input assumptions remain unverified. These scores therefore describe particular paths; they do not establish performance under a common information set or a general ranking of model classes. Unverified historical scores are confined to an appendix. The contribution is a reproducible error audit and correction of the interpretation of an extreme forecast. Neither model is retrained, and no new forecasting method or comprehensive comparison is claimed. Architecture ablations, rolling-origin re-estimation, random-seed sensitivity, and predictive-accuracy inference remain future research. Because the information cutoff of the recorded paths is undocumented, the direction of the informational difference between them and the reconstructed benchmarks cannot be signed.

Authors

Institutions

Publication Details

Journal
Computation
Published
2026-11-16
DOI
https://doi.org/10.3390/computation14110293
Primary Topic
Forecasting Techniques and Applications
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A Retrospective Forecast Audit of Recorded LSTM and ARIMA Inflation Projection Paths for Vietnam, October 2024 to September 2025

The-Ngan Ma, Nhung Thi Nguyen, Huy Khanh Nguyen, Vinh Xuan Hoang et al.
1 links
Computation
Forecasting Techniques and Applications
article

A Retrospective Forecast Audit of Recorded LSTM and ARIMA Inflation Projection Paths for Vietnam, October 2024 to September 2025

The-Ngan Ma, Nhung Thi Nguyen, Huy Khanh Nguyen, Vinh Xuan Hoang, Hieu Trung Le, Phuong Hoang Trinh
article en

Abstract

This study audits the recorded twelve-month LSTM and ARIMA paths for year-over-year consumer price inflation in Vietnam over October 2024–September 2025. Fixed projections are compared with official outcomes and with persistence and seasonal-naive benchmarks reconstructed from observations available after the release on 6 October 2024. LSTM has a path RMSE of 5.524 percentage points, compared with 1.217 for ARIMA, 0.863 for seasonal naive, and 0.596 for persistence. The January 2025 LSTM projection of 20.66% contrasts with the realized inflation of 3.63% and contributes 79.22% of its total squared error. Removing that month in a secondary influence check leaves the descriptive ranking unchanged. Original LSTM and ARIMA issue dates and future-input assumptions remain unverified. These scores therefore describe particular paths; they do not establish performance under a common information set or a general ranking of model classes. Unverified historical scores are confined to an appendix. The contribution is a reproducible error audit and correction of the interpretation of an extreme forecast. Neither model is retrained, and no new forecasting method or comprehensive comparison is claimed. Architecture ablations, rolling-origin re-estimation, random-seed sensitivity, and predictive-accuracy inference remain future research. Because the information cutoff of the recorded paths is undocumented, the direction of the informational difference between them and the reconstructed benchmarks cannot be signed.

ComputationVol. 14(11)
Vietnam National University, Hanoi (VN), VNU University of Economics and Business (VN)
Forecasting Techniques and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon