Paper: arXiv 2308.11939
Authors: Md Sabbirul Haque, Md Shahedul Amin, Jonayet Miah
Abstract
Accurate demand forecasting in the retail industry is a critical determinant of financial performance and supply chain efficiency. As global markets become increasingly interconnected, businesses are turning towards advanced prediction models to gain a competitive edge. However, existing literature mostly focuses on historical sales data and ignores the vital influence of macroeconomic conditions on consumer spending behavior. In this study, we bridge this gap by enriching time series data of customer demand with macroeconomic variables, such as the Consumer Price Index (CPI), Index of Consumer Sentiment (ICS), and unemployment rates. Leveraging this comprehensive dataset, we develop and compare various regression and machine learning models to predict retail demand accurately.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 7.2/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced machine learning models and statistical methods like LSTM and SARIMA, indicating moderate to high math complexity, and uses real-world macroeconomic datasets with comparative model evaluations, demonstrating strong empirical rigor.
Research Flowchart
flowchart TD
A["Research Goal: Accurate retail demand forecasting with macroeconomic variables"] --> B["Data Collection & Integration"]
B --> C{"Data Inputs"}
C --> C1["Historical Sales Data"]
C --> C2["Macroeconomic Indicators<br>CPI, ICS, Unemployment"]
B --> D["Model Development & Comparison"]
D --> E["Computational Process"]
E --> E1["Regression Models"]
E --> E2["Machine Learning Models"]
E --> E3["Forecasting Execution"]
E --> E4["Performance Evaluation<br>MAE, RMSE"]
E4 --> F["Key Findings"]
F --> F1["Macroeconomic variables significantly improve forecast accuracy"]
F --> F2["ML models outperform traditional regression"]
F --> F3["Integrated approach essential for supply chain efficiency"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes enriching retail demand time series with macroeconomic variables and comparing regression and machine learning models for improved forecasting.
Source sentence
we bridge this gap by enriching time series data of customer demand with macroeconomic variables, such as the Consumer Price Index (CPI), Index of Consumer Sentiment (ICS), and unemployment rates.
Assumes 4
- The Walmart sales data is representative of retail demand for the studied products and stores. data
Source sentence
This dataset spans a temporal expanse of five years and encompasses 3,049 distinct products retailed across ten different stores situated within three states: CA, TX and WI.
- Macroeconomic variables from external sources are relevant and correctly aligned with the sales data. data
Source sentence
These economic variables are procured from reputable sources, such as the World Bank's World Development Indicators (WDI) database and the University of Michigan's repository.
- Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) are appropriate metrics for evaluating forecast performance. model
Source sentence
Model performances are measured in terms of Root Mean Squared Error (RMSE) as well as in terms of Mean Absolute Error (MAE).
- Including macroeconomic features will not introduce excessive noise that outweighs their predictive benefit. model
Source sentence
However, simply merging all available data can introduce noise or outdated historical signals.
Data
- Walmart USA retail sales dataset other public
Source sentence
This dataset spans a temporal expanse of five years and encompasses 3,049 distinct products retailed across ten different stores situated within three states: CA, TX and WI.
- Consumer Price Index (CPI) macro public
Source sentence
These economic variables are procured from reputable sources, such as the World Bank's World Development Indicators (WDI) database
- Index of Consumer Sentiment (ICS) macro public
Source sentence
and the University of Michigan's repository.
- Unemployment rates macro public
Source sentence
These economic variables are procured from reputable sources, such as the World Bank's World Development Indicators (WDI) database
Costs
Not applicable
Finds 4
- Incorporating macroeconomic variables yields modest improvements in forecast accuracy for most machine learning models, with LightGBM benefiting the most. predictive-accuracy
Source sentence
Except a few cases, these results in general demonstrate that incorporating external macroeconomic data as model inputs provided mostly modest gains in predictive accuracy across the machine learning models.
- LightGBM MAE decreased from 0.849 to 0.847 and RMSE from 1.717 to 1.715 when macro variables were added. performance
Source sentence
For LightGBM, MAE decreases from 0.849 to 0.847 while RMSE declines from 1.717 to 1.715 when macro variables are added.
- Lasso and Ridge models show only slight improvements or negligible changes with macro variables. performance
Source sentence
The Lasso model obtains a MAE of 0.886 and RMSE of 1.802 without macro variables. Incorporating macroeconomic features slightly improved performance to MAE of 0.884 and RMSE value of 1.798.
- Decision Tree model performance slightly worsens when macro variables are included. negative
Source sentence
the Decision Tree model exhibits a slight rise in errors, with MAE risingg from 1.000 to 1.001 and RMSE rising from 2.357 to 2.364 with macro features.
Checked by others
Supported by
- Ultimate Forward Rate Prediction and its Application to Bond Yield Forecasting: A Machine Learning Perspective
Compare the findings
This paper: Incorporating macroeconomic variables yields modest improvements in forecast accuracy for most machine learning models, with LightGBM benefiting the most.
That paper: Incorporating macroeconomic variables, especially price‑index related ones, significantly improves prediction accuracy.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.