Group12 proposal
Contents
Abstract
Global warming is an ongoing problem, and for a long time scientists have sounded the alarm about taking action to slow it down. Meanwhile, the situation continues to deteriorate. Rising temperatures have led to rising sea levels, changing rainfall patterns and more frequent extreme weather. Recent severe fires in Australia and locust plagues in Africa have been linked to droughts caused by warming temperatures.
It is well known that the main cause of global warming is the massive use of fossil fuels (such as coal, oil, etc.) for nearly a century, which emits a large amount of CO2 and other greenhouse gases. These greenhouse gases absorb infrared light from the radiation emitted by the earth, causing the earth's temperature to rise. This project aims to let everyone know the trend of global warming and the emission of carbon dioxide in various countries, so that everyone can attach importance to this problem.
Objective
The first objective is to forecast the global emissions of carbon dioxide in the next 10 years and use the obtained result as a variable to forecast the future global temperature as well.
The second objective is to visualize our data by creating:
- World maps to show and compare the regional and annual total CO2 emission and per capita CO2 emission with datasets‘National Data’ and ‘Population Data’.
- Time series line charts to show the historical and future gobal CO2 emission and temperature and how they change during different time period as well with dataset ‘Performance data’.
Performance Data
This dataset is the main dataset: According to our research process, we need to use the factors that affect CO2 emissions, the total amount of CO2, the global average temperature, and the number of melting glaciers. The table below shows the main variables that we will be using for our analysis:
National Data
This dataset is the supporting dataset: This is a table of carbon dioxide emissions by country. Through this table we will show a map of carbon dioxide emissions by country. It also shows the impact of each different fossil fuel in this country separately, while making global comparisons.
Visualization
- Time series: The ‘highcharter’, ‘plotly’, ‘viridis’, ‘scatterplot3d’ and ‘ggplot2’ packages will be employed to create two interactive time series line charts, revealing how the historical and future global CO2 emission and temperature change during different period.
- World map: Same packages as mentioned above will be applied to generate two interactive world maps, showing and comparing the CO2 emission in different country and year respectively.
Methodology
Three different approaches will be utilized to predict the global CO2 emissions and temperature in the next 10 years:
- Holt exponential smoothing: By applying this approach, consequently each relevant variables’ (e.g. gas fuel, liquid fuel and solid fuel) future value will be obtained. And we can use them to predict the future CO2 emission by employing the linear regression model.
- SARIMA: Seasonal Autoregressive Integrated Moving Average (SARIMA) model, an extension of ARIMA that explicitly supports univariate time series data with a seasonal component will be applied to conduct the prediction. We can use it gain the annual CO2 emission in the future with a lower and upper bound.
- Auto-Regression: The Auto-Regression model describes the relationship between current values and the historical values. And it uses the historical time data as the variable to predict its future value. The factors that influence the CO2 emission, such as solid fuel and gas fuel, can be predicted by Auto-Regression model. As a result, the future global CO2 emission will be predicted by employing the linear regression model.
After completing all prediction methods mentioned above, we intend to compare their result respectively with the actual CO2 emission in recent years as an evaluation and determine which of them is the best fit one.
Critics of Existing Works
- Exploring the CO2 and GHG data
https://www.kaggle.com/martijnjschut/exploring-the-co2-and-ghg-data
This notebook firstly reviews the history data of the global CO2 emission from 1751 to 2017 and visualizes it with a line chart by applying the time series analysis revealed how the global CO2 emission changed during different time period. Then, a world map was created to indicate the CO2 emission in each country. And audiences can compare the values between different country or continent by easily observing their color. Even though the visualization has successfully showed a plenty of important information that author trying to tell, there are still some drawbacks. For example, both graphs have no unit of measurement, which makes operators difficult to know the accurate implication of axis labels. And the lack of title and annotation might make people hard to catch the most important aspects of the chart and identify what is relevant. Besides, the world map is not completely interactive to user. Audiences are not free to view the regional CO2 emission in any single year they want but in year 2017 only.
- CO2 Emission Forecast
https://www.kaggle.com/berhag/co2-emission-forecast-with-python-seasonal-arima
After diagnosing and validating, this notebook forecasts the future global CO2 emission caused by the natural gas consumption in the future 10 years since 2016 by employing the Seasonal Autoregressive Integrated Moving Average (SARIMA) model with grid search method. Consequently, a time series line graph is generated, and readers can observe the next ten years’ predicted CO2 emissions with lower and upper bound. Generally, the whole visualization has been done extremely well, but it lacks some clarity such as the missing title and unit of measurement. In terms of aesthetics, the color coding is not very distinguishable, and the graph can be modified longer to show the change of line more clearly. On the other hand, as a further extension, we are going to use those values to forecast the future global temperature, which will be applied to obtain the anticipated sea level eventually. And an extra time series line graph will be created to reveal how exactly it changes.
Data Scource
The data from the “Climate Change Data” and “Population, total”.
Team Members
- Lyu Ruiqi
- Yao Zhiqian
- Zhao Chenyang
