In this study, I explore the application of the Autoregressive Integrated Moving Average (ARIMA) model to forecast electric SUV sales, leveraging historical data from 2006 to 2024. The primary objective is to analyze trends and predict future sales, thereby supporting strategic decision-making in the automotive industry and contributing to the development of new quality productivity. As the global shift toward digital and green economies accelerates, the electric SUV sector has demonstrated robust growth, playing a pivotal role in modern industrial systems. By employing time series analysis, I aim to provide insights that enhance market strategies, policy formulation, and investment decisions, ultimately fostering innovation and sustainability in the electric SUV market.
The research is grounded in the increasing popularity of electric SUV models, which offer advantages such as spacious interiors, high off-road capability, and superior driving experiences. However, the volatility and unpredictability of the automotive market necessitate accurate sales forecasting. Recent developments, including the successful launch of models like the Xiaomi SU7, have intensified competition and spurred upgrades in the electric SUV segment. These vehicles excel in areas like intelligence and connectivity, making them a key growth driver in the automotive industry. Forecasting electric SUV sales is crucial for manufacturers, policymakers, and investors to navigate market dynamics effectively.
Sales predictions for electric SUV units enable stakeholders to align production, marketing, and policy efforts with consumer demand. For instance, automotive companies can optimize inventory management and resource allocation, while governments can design incentives and regulations that promote low-carbon technologies. Moreover, the entire supply chain—from component suppliers to dealerships—benefits from anticipating market shifts. Theoretically, this study advances the application of time series analysis in economics and management, demonstrating its utility in real-world scenarios. From an investment perspective, accurate forecasts help identify growth opportunities and mitigate risks in the electric SUV market.
To model the sales data, I utilize the ARIMA framework, which combines autoregressive (AR), differencing (I), and moving average (MA) components. The model is denoted as ARIMA(p, d, q), where p represents the order of the AR term, d the degree of differencing to achieve stationarity, and q the order of the MA term. For a time series \(X_t\), the ARIMA model can be expressed as:
$$ \phi(B)(1-B)^d X_t = \theta(B) \varepsilon_t $$
where \(\phi(B)\) and \(\theta(B)\) are polynomial functions of the backshift operator B, and \(\varepsilon_t\) is the error term. In cases with seasonal patterns, the model extends to ARIMA(p, d, q)(P, D, Q)[S], incorporating seasonal parameters. To generate forecasts, I employ a bootstrapping approach with 1000 iterations, simulating residuals and seasonal components to project future electric SUV sales. This method enhances prediction reliability by accounting for uncertainty in the data.
The dataset comprises monthly electric SUV sales figures from May 2006 to January 2024, totaling 213 observations. Initial visualization revealed a clear upward trend with significant seasonal fluctuations, particularly in recent years. To address non-stationarity, I applied logarithmic transformation followed by first-order seasonal differencing (D=1, S=12) and first-order ordinary differencing (d=1). This preprocessing step reduced volatility and improved the series’ stability, as confirmed by subsequent tests. The transformed data exhibited less erratic behavior, making it suitable for ARIMA modeling. The electric SUV sales trajectory underscores the segment’s resilience and potential for continued expansion.
To assess stationarity, I conducted the Augmented Dickey-Fuller (ADF) test, which checks for unit roots in the time series. The null hypothesis posits that the data is non-stationary. For the original electric SUV sales series, the ADF statistic was 0.4234 with a p-value of 0.9823, failing to reject the null hypothesis and indicating non-stationarity. After differencing, the ADF statistic improved to -5.009858 with a p-value of 0.00954, allowing rejection of the null hypothesis and confirming stationarity. A white noise test on the differenced series confirmed that it contained meaningful patterns, justifying further analysis. The results are summarized in Table 1 below.
| Data Type | ADF Statistic | P-value | Stationarity |
|---|---|---|---|
| Original Series | 0.4234 | 0.9823 | Non-stationary |
| Differenced Series | -5.009858 | 0.00954 | Stationary |
With stationarity established, I proceeded to model identification and diagnostics. Using grid search, I evaluated ARIMA parameters from (0,1,0)(0,1,0)[12] to (3,1,3)(2,1,2)[12], focusing on minimizing the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). The RMSE and MAE are defined as:
$$ \text{RMSE} = \sqrt{\frac{1}{n} \sum_{i=1}^n (Y_i – \hat{Y}_i)^2} $$
$$ \text{MAE} = \frac{1}{n} \sum_{i=1}^n |Y_i – \hat{Y}_i| $$
where \(Y_i\) represents actual electric SUV sales, \(\hat{Y}_i\) denotes predicted values, and n is the number of observations. The optimal model, ARIMA(3,1,0)(0,1,1)[12], achieved the lowest RMSE (110087.49) and MAE (67272.50). Residual diagnostics confirmed that the model residuals were white noise, with significant coefficients (p-values < 0.05), ensuring robust predictions. Table 2 compares the performance metrics across select models.
| ARIMA Model | RMSE | MAE |
|---|---|---|
| (1,1,1)(1,1,1)[12] | 120456.78 | 74532.10 |
| (2,1,1)(0,1,1)[12] | 115632.45 | 70341.25 |
| (3,1,0)(0,1,1)[12] | 110087.49 | 67272.50 |
| (2,1,2)(1,1,1)[12] | 118765.33 | 72891.60 |
Leveraging the ARIMA(3,1,0)(0,1,1)[12] model, I generated half-bootstrapping forecasts for electric SUV sales from February to June 2024. The predictions indicate a gradual upward trend, with sales expected to surpass 1.78 million units by May 2024. Table 3 presents the forecasted values, including mean estimates and 95% confidence intervals, derived from 1000 bootstrap samples. This approach accounts for variability and provides a range of possible outcomes, enhancing decision-making reliability.
| Month | Mean Forecast | 2.5% Lower Bound | 97.5% Upper Bound |
|---|---|---|---|
| 2024-02 | 1,566,802 | 1,136,474.5 | 2,194,091 |
| 2024-03 | 1,393,084 | 912,510.9 | 2,101,028 |
| 2024-04 | 1,589,573 | 984,708.7 | 2,562,212 |
| 2024-05 | 1,786,506 | 947,933.8 | 2,817,867 |
| 2024-06 | 1,642,797 | 737,171.4 | 2,721,865 |
The forecasting results suggest that electric SUV sales will experience minor fluctuations but overall growth in the coming months. However, real-world factors such as economic conditions, policy changes, and seasonal events—like the Chinese New Year—can cause deviations from model predictions. For instance, actual sales in February 2024 may dip due to holidays, highlighting the need to complement quantitative forecasts with qualitative insights. Despite these nuances, the electric SUV market remains a powerhouse, driven by consumer preference for customized, intelligent vehicles and supportive environmental policies.
In conclusion, this study demonstrates the efficacy of ARIMA models in predicting electric SUV sales, offering valuable tools for industry stakeholders. The upward trajectory of electric SUV adoption aligns with broader trends in digitalization and sustainability, reinforcing its role in advancing new quality productivity. By integrating data-driven forecasts into strategic planning, businesses and governments can optimize resources, stimulate innovation, and reduce carbon emissions. Future research could incorporate external variables, such as economic indicators or technological advancements, to refine predictions further. Ultimately, the growth of the electric SUV sector signifies a transformative shift toward smarter, greener mobility solutions.
The implications of this analysis extend beyond sales numbers. For automotive manufacturers, forecasting electric SUV demand enables efficient production scheduling and inventory management, reducing costs and enhancing competitiveness. Policymakers can use these insights to design incentives for electric vehicle adoption, infrastructure development, and R&D funding. Investors gain a clearer picture of market potential, guiding capital allocation toward high-growth areas like electric SUV technologies. Moreover, the emphasis on electric SUV models underscores their contribution to ecological modernization and energy efficiency, key components of new quality productivity. As the industry evolves, continuous monitoring and model updates will be essential to capture emerging trends and disruptions.
From a methodological perspective, the ARIMA framework proves highly adaptable for electric SUV sales forecasting, but it has limitations, such as assuming linear patterns and overlooking external shocks. Future work could explore hybrid models combining ARIMA with machine learning techniques to improve accuracy. Additionally, expanding the dataset to include global electric SUV sales could provide a more comprehensive view. Nonetheless, the current findings offer a solid foundation for understanding market dynamics and fostering sustainable growth in the electric SUV ecosystem. By prioritizing data analytics and innovation, stakeholders can harness the full potential of electric SUV to drive economic and environmental progress.