Introduction
Project Inspiration
It is obvious to anyone that seasonal changes work hand-in-hand with tourism. With Malta being known for its hot sunny climate, it makes for an ideal destination spot for anyone looking for deep-rooted history and culture, traditional architecture, as well as its beautiful beaches.
However, it can also be noted that within the last decade, the number of tourists every summer has sky rocketed. Therefore, I wanted to know exactly WHY this increase has occurred. Is it because of Malta making a bigger name for itself internationally? Or is there another factor?
Upon thinking on this, as a Maltese citizen, it is clear that our summers have gotten drier and hotter, and rather than this deterring tourists, it has brought in an influx of them. Therefore, I wanted concrete correlative data finding the link between these weather conditions and tourists, and whether or not they have as much of an impact as I thought.
mean_temp max_temp feels_like sunshine_dur daylight_dur \
0 17.157497 18.857500 12.822860 34635.476562 35184.191406
1 15.365833 16.007500 7.778753 34146.585938 35220.207031
2 14.561668 15.107500 8.405110 33864.101562 35259.179688
3 14.834582 16.807501 11.814759 27727.416016 35301.085938
4 16.895000 18.107500 12.998100 30295.017578 35345.867188
... ... ... ... ... ...
5839 15.510418 16.700001 12.891060 32277.310547 35051.183594
5840 14.535417 17.100000 13.406756 33581.242188 35071.515625
5841 13.752086 16.350000 13.152936 29120.347656 35094.996094
5842 13.564583 15.400000 11.221592 33344.414062 35121.636719
5843 13.343750 15.400000 10.727241 32398.789062 35151.402344
precipitation wind_speed weather_code
0 0.0 43.601631 3.0
1 0.2 54.553955 51.0
2 0.0 45.424240 2.0
3 0.0 29.024128 3.0
4 0.0 36.546650 3.0
... ... ... ...
5839 1.3 26.221373 53.0
5840 0.1 19.361507 51.0
5841 1.4 12.749540 53.0
5842 1.2 25.264202 53.0
5843 1.7 29.606031 53.0
[5844 rows x 8 columns]
freq c_resid unit nace_r2 geo\TIME_PERIOD 1990-01 1990-02 1990-03 \
1469 M TOTAL NR I551-I553 MT NaN NaN NaN
1990-04 1990-05 ... 2025-05 2025-06 2025-07 2025-08 2025-09 \
1469 NaN NaN ... 258737.0 252970.0 265532.0 271574.0 245468.0
2025-10 2025-11 2025-12 2026-01 2026-02
1469 283159.0 225437.0 205869.0 174318.0 NaN
[1 rows x 439 columns]
mean_temp max_temp feels_like sunshine_dur daylight_dur \
2010-01 13.8 18.9 8.9 260.265167 311.221917
2010-02 13.8 19.0 9.3 261.864583 304.836944
2010-03 14.4 18.3 11.1 315.765556 371.162278
2010-04 16.1 19.4 14.2 352.021778 393.274111
2010-05 18.5 24.2 16.6 385.540556 436.067917
... ... ... ... ... ...
2025-08 26.5 32.3 29.6 388.778278 417.751806
2025-09 25.4 32.8 27.9 342.742056 371.764194
2025-10 21.4 26.9 21.3 309.080611 349.124889
2025-11 18.2 24.2 17.0 257.064333 308.276250
2025-12 15.3 20.4 13.7 247.401250 302.845306
precipitation wind_speed weather_code weather_desc
2010-01 56.8 41.2 3 Overcast
2010-02 22.9 40.3 3 Overcast
2010-03 41.7 33.9 3 Overcast
2010-04 11.0 29.0 3 Overcast
2010-05 2.9 31.7 3 Overcast
... ... ... ... ...
2025-08 1.0 16.0 0 Clear sky
2025-09 26.7 15.2 3 Overcast
2025-10 51.0 21.6 3 Overcast
2025-11 34.4 21.0 51 Drizzle: Light
2025-12 82.9 19.7 3 Overcast
[192 rows x 9 columns]
tourist_arrivals 2010-01 49634 2010-02 65960 2010-03 84836 2010-04 104939 2010-05 123745 ... ... 2025-08 271574 2025-09 245468 2025-10 283159 2025-11 225437 2025-12 205869 [192 rows x 1 columns]
tourist_arrivals mean_temp max_temp feels_like sunshine_dur \
2010-01 49634 13.8 18.9 8.9 260.265167
2010-02 65960 13.8 19.0 9.3 261.864583
2010-03 84836 14.4 18.3 11.1 315.765556
2010-04 104939 16.1 19.4 14.2 352.021778
2010-05 123745 18.5 24.2 16.6 385.540556
... ... ... ... ... ...
2025-08 271574 26.5 32.3 29.6 388.778278
2025-09 245468 25.4 32.8 27.9 342.742056
2025-10 283159 21.4 26.9 21.3 309.080611
2025-11 225437 18.2 24.2 17.0 257.064333
2025-12 205869 15.3 20.4 13.7 247.401250
daylight_dur precipitation wind_speed weather_code weather_desc
2010-01 311.221917 56.8 41.2 3 Overcast
2010-02 304.836944 22.9 40.3 3 Overcast
2010-03 371.162278 41.7 33.9 3 Overcast
2010-04 393.274111 11.0 29.0 3 Overcast
2010-05 436.067917 2.9 31.7 3 Overcast
... ... ... ... ... ...
2025-08 417.751806 1.0 16.0 0 Clear sky
2025-09 371.764194 26.7 15.2 3 Overcast
2025-10 349.124889 51.0 21.6 3 Overcast
2025-11 308.276250 34.4 21.0 51 Drizzle: Light
2025-12 302.845306 82.9 19.7 3 Overcast
[192 rows x 10 columns]
=== CORRELATION RESULTS ===
max_temp 0.545253
feels_like 0.538703
mean_temp 0.521722
daylight_dur 0.412457
sunshine_dur 0.389461
precipitation -0.219104
weather_code -0.219966
wind_speed -0.555857
Name: tourist_arrivals, dtype: float64
=== LASSO REGRESSION RESULTS ===
feature coefficient
0 mean_temp 9905.283970
2 daylight_dur 5982.507408
3 precipitation 2037.868551
1 sunshine_dur -0.000000
5 weather_code -226.074359
4 wind_speed -20341.859643
sunshine_dur has been dropped by Lasso.
Precipitation Stats
count 192.000000
mean 32.660937
std 34.849282
min 0.000000
25% 3.300000
50% 23.350000
75% 50.525000
max 164.900000
Name: precipitation, dtype: float64
Result Analysis
Positive Trend Indicators
- mean_temp
The strongest positive trend indicator was the 'mean_temp' feature, showing that approximately 10,000 more tourists come to Malta when temperatures are higher.
- daylight_dur
The second strongest positive trend indicator was 'daylight_dur'. This shows that the duration of daylight is more influencial than the duration of sunshine.
- precipitation
A surprising 'positive' trend indicator. This was deemed anomolous due to the illogical coefficient (more rain bringing more tourists) and its weak influence.
After analysing the stats of precipitation, it was determined that there was not enough variance to determine a reasonable positive influence, which is why precipitation was dropped.
Impactless Features
'sunshine_dur' and 'weather_code' were both dropped as the Lasso Regression model determined that neither of these features influence the number of tourists.
Negative Trend Indicators
1. wind_speedWhile being the only negative indicator, it had the largest influence on tourists. Approximately 20,000 less tourists visited Malta when weather featured strong winds.
Result Analysis
Noticeable Considerations
2020 - 2022 Tourism Dip
Data from the 2020-2022 period was excluded from this visualisation to account for the tourism dip due to the COVID-19 pandemic. Removing this anomaly ensures that the visualisation remains consistent with long-term trends and consumer sentiment.
Expected Results
mean_temp and daylight_hours both provided the expected results. During colder months containing less daylight hours, tourism has a noticeable dip. This aligns with the preconceived notion of tourism peaking during the summer. Both of these features move proportionally with the Tourist Arrivals, confirming the positive coefficient made by the Lasso Regression model earlier on.
Unexpected Results
Alternatively, wind_speed shows a decline over this time period, sparking doubts regarding potential data inaccuracies. As Malta becomes an increasingly more popular tourist attraction over the years, there are 2 possible scenarios to be looked into:
- Data Measurement Inaccuracies
This scenario implies that either changes in measurement methodology or data source rather than a genuine meteorological trend, supported by the structural break in between 2016 and 2018, which clearly shows that this is not a gradual decline caused by climate change. Furthermore, implying that this extreme negative correlation trend is purely coincidental (as wind speed decreased due to these inaccuracies, Malta's popularity rose - displaying an inversely proportional relationship).
- Underlying Correlation
This scenario is strengthened by the Lasso Regression Coefficient of -20,418, the strongest coefficient in the model, suggesting that statistically, wind speed has a strong statisitical influence on tourist_arrivals. Hence, it shows that wind_speed is a feature that is often overlooked, indicating that the decreasing wind speeds made Malta more attractive for tourists to visit.
malta travel 2010-01 46 2010-02 31 2010-03 42 2010-04 55 2010-05 42 ... ... 2025-08 34 2025-09 22 2025-10 23 2025-11 27 2025-12 33 [192 rows x 1 columns]
Result Analysis
It can be observed that the Search Interest follows the trend of Number of Tourists. Notably, search interest begins peaking slightly before the start of the hot weather, suggesting that the intent to travel comes before the actual tourist season. This reinforces the notion that weather is a direct driver of tourism demand.
However, Search Interest remains relatively consistent throughout the timeframe, with no noticeable long-term surges. This suggests that it captures seasonal patterns, rather than growing popularity.
This pattern remains consistent across both the temperature and daylight duration, both of which increase proportionally with the search interest. Alternatively, wind speed moves inversely to the search interest, which remains consistent with the negative correlation identified in earlier findings.
Conclusions
Overall, the analysis suggests that weather conditions, particularly temperature and daylight hours, play a significant role in Malta's tourism patterns, with warmer sunnier months consistently attracting higher visitor numbers. This is further reinforced by Google Trends data, which shows search interest in Malta travel peaking slightly before the start of the warmer weather, suggesting that weather anticipation is a direct contributing factor during travel planning. The features temperature and daylight hours show consistent recordings throughout the years, and clear and consistent correlation patterns relative to the number of tourists. On the other hand, wind speed has the potential to be a deciding factor in tourist arrivals if and only if the data can be confirmed to be accurate.