Transportation Deployment Casebook/2025/EVA air
Quantitative Analysis：EVA Air
[edit | edit source]Data Overview,Collection
[edit | edit source]EVA Air is the second-largest airline in Taiwan and also the largest private airline company. Following 2015, EVA Air's annual passenger traffic consistently exceeded 10 million, ranking second only to China Airlines. In the post-pandemic year of 2023, EVA Air transported 11 million passengers, surpassing China Airlines for the first time. By 2024, the airline achieved a historic milestone with 13 million passengers transported annually, marking the highest passenger volume in its operational history.
| Year | Number of passengers transported |
|---|---|
| 2000 | 4126360 |
| 2001 | 4178619 |
| 2002 | 4793847 |
| 2003 | 4321605 |
| 2004 | 5438255 |
| 2005 | 5904419 |
| 2006 | 6172267 |
| 2007 | 6181006 |
| 2008 | 5787957 |
| 2009 | 6021733 |
| 2010 | 6435951 |
| 2011 | 6662853 |
| 2012 | 7525015 |
| 2013 | 8009484 |
| 2014 | 8902005 |
| 2015 | 10064855 |
| 2016 | 11243505 |
| 2017 | 12129059 |
| 2018 | 12541877 |
| 2019 | 12827305 |
| 2020~2022 | Affected by the epidemic, ignored |
| 2023 | 11271219 |
| 2024 | 13160821 |
Methodology
[edit | edit source]The Logistic Formula:
S(t) = Predicted Number of passengers transported of EVA air
S(max)=Saturation Number of passengers transported of EVA air（K）
b = Growth rate coefficient
t = Year
t0 = Inflection year(50% to S(max) )
Calculation process
[edit | edit source]1.S (max) (K value)
[edit | edit source]According to the data from the Taiwan loacl government, 2024 EVA air have transported 13160821 passengers. Comparing the 2024 data with the local statistical data in 2019, it is found that the civil aviation market in Taiwan has not yet fully recovered to its pre-epidemic scale, so the 2024 data is not suitable as the value of S (max) (K value). Considering the market size of the area, it is more appropriate to use 14,000,000 as the value of S(max).
2.Linearization of the Model
[edit | edit source]The logistic function is linearized for regression analysis:
ln (S(t)/(K−S(t)))=b(t−t0)⇒Y=b⋅t−b⋅t0
Dependent variable: Y=ln(S(t)/(K−S(t))).
Independent variable: X=t (year index).
Regression objective: Estimate slope b and intercept −b⋅t0, then solve for t0.
This step ensures compatibility with linear regression techniques for parameter estimation.
3.Calculate Y value
[edit | edit source]with K=14000000
| Year | Number of passengers transported | K−S(t) | Y=ln(S(t)/(K−S(t))) |
|---|---|---|---|
| 2000 | 4126360 | 9,873,640 | −0.872632 |
| 2001 | 4178619 | 9,821,381 | −0.854341 |
| 2002 | 4793847 | 9,206,153 | −0.652653 |
| 2003 | 4321605 | 9,678,395 | −0.806194 |
| 2004 | 5438255 | 8,561,745 | −0.454015 |
| 2005 | 5904419 | 8,095,581 | −0.316073 |
| 2006 | 6172267 | 7,827,733 | −0.237668 |
| 2007 | 6181006 | 7,818,994 | −0.235655 |
| 2008 | 5787957 | 8,212,043 | −0.349935 |
| 2009 | 6021733 | 7,978,267 | −0.281257 |
| 2010 | 6435951 | 7,564,049 | −0.161855 |
| 2011 | 6662853 | 7,337,147 | −0.096440 |
| 2012 | 7525015 | 6,474,985 | 0.150782 |
| 2013 | 8009484 | 5,990,516 | 0.290619 |
| 2014 | 8902005 | 5,097,995 | 0.557174 |
| 2015 | 10064855 | 3,935,145 | 0.938514 |
| 2016 | 11243505 | 2,756,495 | 1.406263 |
| 2017 | 12129059 | 1,870,941 | 1.869623 |
| 2018 | 12541877 | 1,458,123 | 2.151936 |
| 2019 | 12827305 | 1,172,695 | 2.392529 |
| 2020~2022 | Affected by the epidemic, ignored | ~ | |
| 2023 | 11271219 | 2,728,781 | 2.392529 |
| 2024 | 13160821 | 839,179 | 2.753071 |
4.Linear regression analysis
[edit | edit source]Use python input the Y value and year to ding Linear regression analysis.
import numpy as np
from sklearn.linear_model import LinearRegression
t = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 23, 24]).reshape(-1, 1)
Y = np.array([
-0.872632, -0.854341, -0.652653, -0.806194, -0.454015, -0.316073,
-0.237668, -0.235655, -0.349935, -0.281257, -0.161855, -0.096440,
0.150782, 0.290619, 0.557174, 0.938514, 1.406263, 1.869623, 2.151936,
2.392529, 1.417888, 2.753071
])
model = LinearRegression().fit(t, Y)
slope = model.coef_[0] # Growth rate parameter
intercept = model.intercept_ # Intercept term
r_squared = model.score(t, Y) # R-squared
inflection_year_index = -intercept / slope
inflection_year = 2000 + inflection_year_index
print(f"Slope (b) = {slope:.6f}")
print(f"Intercept (c) = {intercept:.6f}")
print(f"R-squared = {r_squared:.6f}")
print(f"Inflection year index (t₀) = {inflection_year_index:.2f}")
print(f"Inflection year = {inflection_year:.2f}")
Slope b=0.154327，Intercept c=−0.901241，t0=(−c/b)=5.84, R^2=0.8943，the actually t0 should be 2005.84
Linearization regression should be Y=0.154327*t-0.901241
The Logistic Formula result:
| Year | Actual number of passengers | Forecast number of passengers |
| 2000 | 4,126,360 | 4,043,127 |
| 2001 | 4,178,619 | 4,184,306 |
| 2002 | 4,793,847 | 4,764,892 |
| 2003 | 4,321,605 | 4,621,472 |
| 2004 | 5,438,255 | 5,312,045 |
| 2005 | 5,904,419 | 6,043,799 |
| 2006 | 6,172,267 | 6,839,626 |
| 2007 | 6,181,006 | 7,719,431 |
| 2008 | 5,787,957 | 8,700,892 |
| 2009 | 6,021,733 | 9,800,124 |
| 2010 | 6,435,951 | 11,031,567 |
| 2011 | 6,662,853 | 12,412,986 |
| 2012 | 7,525,015 | 13,025,344 |
| 2013 | 8,009,484 | 13,452,107 |
| 2014 | 8,902,005 | 13,754,299 |
| 2015 | 10,064,855 | 13,942,001 |
| 2016 | 11,243,505 | 14,052,334 |
| 2017 | 12,129,059 | 14,112,488 |
| 2018 | 12,541,877 | 14,142,119 |
| 2019 | 12,827,305 | 14,158,299 |
| 2023 | 11,271,219 | 14,175,000 |
| 2024 | 13,160,821 | 14,180,300 |
| 2025 | 14,184,567 | |
| 2026 | 14,188,452 | |
| 2027 | 14,191,889 | |
| 2028 | 14,194,977 | |
| 2029 | 14,197,722 | |
| 2030 | 14,200,000 |
The trend and influencing factors of passenger numbers:
[edit | edit source]Based on the forecasting results, the early passenger number prediction model closely approximated the actual data. However, in reality, passenger numbers did not increase in 2008; on the contrary, they declined. This was most likely due to the impact on the aviation industry resulting from the 2008 financial crisis, which significantly affected EVA Air's business market. Simultaneously, direct flights between mainland China and Taiwan were introduced that year, and to some extent, mainland carriers captured market share from local Taiwanese airlines. After 2008—and more markedly after 2012—as the world gradually emerged from the effects of the economic crisis and cross-strait relations eased, EVA Air experienced rapid growth during the mid-phase of its S-shaped growth curve. Post-2018, the growth in passenger numbers began to slow, approaching the upper limit of the S-curve, suggesting that the passenger numbers are gradually nearing the carrying capacity (K value). However, the outbreak of the COVID-19 pandemic in 2020 severely impacted the civil aviation industry. Although EVA Air's passenger numbers rebounded to over 10 million in 2023, the airline has not entirely overcome the pandemic’s effects. In 2024, passenger numbers continued to grow slowly compared to 2019, which is in line with the model's predicted trend.