All countries showed an increase in temperatute.
Global warming seems to be affecting more the countries in the north. The highest increases seem to be in Asia, Europe and Canada.
Countries around the Equator and on the Southern Hemisphere have only moderate temperature increases.
May 14, 2018
Introduction
Main findings
Model Explanation
Seasonal Pattern - Example Switzerland
- In the model we used month and year
- Month because we expect countries to have a seasonal pattern
- Year because it captures the overtime increase in temperature
Model Fit
World map yearly increase in temperature
Model quality
Model Quality
The model quality depends on the existence of clear seasonal pattern.
Countries without a strong seasonal pattern the model finds hard to predict the monthly averages
Countries in Europe and Asia have a strong seasonal pattern
Next we compare the data of the country with the worst fit (Rwanda), with the one with the best fit (China), were it will become clear what causes the difference in model quality.
China
- The strong, stable seasonal pattern in China's recorded temperature data allows for an accurate prediction of monthly temperature averages
Rwanda
- For Rwanda, however, the model is unable to learn well from past information, as the seasonal pattern of any year is unlikely to be replicated exactly in the following years.
Unsurprisingly, as the seasonal patterns are the weakest for countries close to the equator, these countries have the lowest model quality. For countries further away from the equator the model seems to work really well, with the majority of the adjusted \(R^2\) values being higher than 0.90.
Interestingly, the countries for which the model has the best fit are all quite large. In fact, within China different areas have different climates. This poses questions as to how the temperature levels are recorded for these countries.