Are we living in a sustainable world? Let’s use Deep Learning (SOM)to find answers.

Sustainable world is the outcome of the sustainable development strategies and implementation. Major factors, such as human population size, biosphere robustness, resource stock, food supply, and environmental quality must remain in balance, on a global scale. This state of balance must last long enough so that it will not be merely a blip on the curve of unsustainable growth . Even though we might not really attain that balance, we must move in that direction if humanity and the ecosystem are to survive.

Because the Earth is a closed system, a sustainable world is not compatible with “sustainable growth” . A closed system might conceivably accommodate “sustainable development,” a term popularized by the World Commission on Environment and Development (1987; the “Brundtland Commission”).

Why Environmental Sustainability?

Environmental sustainability is concerned with whether environmental resources will be protected and maintained for future generations.Environmental sustainability is concerned with issues such as:

• Long-term health of ecosystems. Protecting the long-term productivity and health of resources to meet future economic and social needs, e.g. protecting food supplies, farmland and fishing stocks.
• Intergenerational decision making. When making economic decisions, we should focus on implications for future generations, and not just the present moment. For example, burning coal gives a short-term benefit of cheaper energy, but the extra pollution imposes costs on future generations.
• Renewable resources: Diversifying into energy sources that do not rely on non-renewable resources. For example, solar and wind power.
• Prevent the consequences of man-made global warming. Policies to ensure the environment of the planet does not deteriorate to a point where future generations face water shortages, extreme weather events, excess temperature. — All factors that could make living in parts of the world very difficult if not possible.
• Protection of species diversity and ecological structure. Sometimes medicines require elements within specific plant species. If some species go extinct, it limits future technological innovation.
• Treating environmental resources as if they have intrinsic rights and value. In other words, we shouldn’t just rely on a monetary value, i.e. we should protect rainforests because they deserve to be protected rather than using a cost-benefit analysis of whether we gain financially from protecting rainforests.
• Targetting social welfare, happiness and environmental sustainability above crude measures of progress such as GDP.

Human Development Report Data

United Nations Development Program publishes Human Development Report that collects data from various sources pertaining to Human Development. It captures various parameters such as ‘quality of human development’, ‘gender gap’, ‘women empowerment’, ‘environment sustainability’ and many more. Please check out the latest report here.

Enviormental Performance Index

Source — Environmental Performance Index (EPI) designed to supplement the environmental targets set forth in the United Nations Millennium Development Goals.

The map depicts the environmental performance of the countries on a scale of 0 to 100. We will try to use Unsupervised Deep Learning technique to organise the countries using Human Development Report Data and try to compare the results.

Unsupervised Deep Learning (Self Organising Maps)

A self-organizing map (SOM) is a type of artificial neural network (ANN) that is trained using unsupervised learning to produce a low-dimensional (typically two-dimensional), discretized representation of the input space of the training samples, called a map, and is therefore a method to do dimensionality reduction. Self-organizing maps differ from other artificial neural networks as they apply competitive learning as opposed to error-correction learning (such as backpropagation with gradient descent), and in the sense that they use a neighborhood function to preserve the topological properties of the input space.

The example shows a complex data set consisting of a massive amount of columns and dimensions and demonstrates how that data set’s dimensionality can be reduced.

So, instead of having to deal with hundreds of rows and columns (because who would want that!), the data is processed into a simplified map; that’s what we call a self-organizing map. The map provides you with a two-dimensional representation of the exact same data set; one that is easier to read.

Building our dataset

Most of the data used in the project was from Human Development Report-2018. The dataset was obtained by following method:

• 1: Slicing the relevant data from the ‘Human Development Report-2018’ and converting it to csv file
• 2: Adding the pollution index from here to the csv file using webscrapping tools in pandas.
• 3: Adding country code to the csv file that will be used to plot choropleth map using plotly

Describing the Columns

• Fossil_fuel_usage: Percentage of total energy consumption that comes from fossil fuels, which consist of coal, oil, petroleum and natural gas products.
• Renewable_energy_usage: Share of renewable energy in total final energy consumption. Renewable sources include hydroelectric, geothermal, solar, tides, wind, biomass and biofuels.
• Per_capita_CO2: Human-originated carbon dioxide emissions stemming from the burning of fossil fuels, gas flaring and the production of cement. Carbon dioxide emitted by forest biomass through depletion of forest areas is included. Data are expressed in tonnes per capita (based on midyear population)
• Forest_area: Land spanning more than 0.5 hectare with trees taller than 5 metres and a canopy cover of more than 10 percent or trees able to reach these thresholds in situ. It excludes land predominantly under agricultural or urban land use, tree stands in agricultural production systems (for example, in fruit plantations and agroforestry systems) and trees in urban parks and gardens. Areas under reforestation that have not yet reached but are expected to reach a canopy cover of 10 percent and a tree height of 5 metres are included, as are temporarily unstocked areas resulting from human intervention or natural causes that are expected to regenerate.
• fresh_water_withdrawl: Total fresh water withdrawn, expressed as a percentage of total renewable water resources.
• natural_resource_reduction: Monetary valuation of energy, mineral and forest depletion, expressed as a percentage of gross national income (GNI).
• air_pollution_deaths_per_100000: Age-standardized mortality rate resulting from exposure to ambient (outdoor) air pollution and household (indoor) air pollution from solid fuel use for cooking, expressed per 100,000 population. Ambient air pollution results from emissions from industrial activity, households, cars and trucks.”
• water_related_deaths_per_100000: Mortality rate attributed to unsafe water, sanitation and hygiene services: Deaths attributable to unsafe water, sanitation and hygiene focusing on inadequate wash services, expressed per 100,000 population.
• land_degradation: Rain fed cropland, irrigated cropland, or range, pasture, forest and woodlands that have experienced the reduction or loss of biological or economic productivity and complexity resulting from a combination of pressures, including land use and management practices.
• red_list_index: Measure of the aggregate extinction risk across groups of species. It is based on genuine changes in the number of species in each category of extinction risk on the International Union for Conservation of Nature Red List of Threatened Species. It ranges from 0, all species categorized as extinct, to 1, all species categorized as least concern.
• Pollution Index: Air quality score. Higher the score, worse is the air quality

Building the Self Organising maps model

We will be using the minisom library developed by ‘JustGlowing’

Visualising 2-D matrix

‘SOM’ assigns each row to a node seperated by another node by a certain Euclidean Distance. So, we will plot heatmap of the Euclidean Distances to identify similar and dissimilar nodes. Higher the Euclidean Distance, larger is the dissimilarity between two nodes.

As we can see that the countries are classified and allocated a node on the basis of how similar they are.

Results

1. From Environmental Performance Index:

2. From the model

As we can see that environmental performance is lowest among countries like India, Pakistan, Nigeria, Yemen, Saudi Arabia, Congo and other African countries. Countries like Canada, US, Brazil, Australia and West Europian countries are perfoming well on the sustainability index. We can compare the results wit the Environmental Performance Index we see that our model produces similar results.

Conclusion

India and neighbouring countries such as Pakistan, Bangladesh have huge populations that put pressure on the environment. Also, central and western African countries are know to have largest portion of the natural resources on the planet.

Sustainability is not a choice but a necessity. Governments across the world need to take drastic actions to reverse the environmental damage. If we are living in the countries low on the environmental performance index, we should take measures to lower our carbon footprint at individual level.

If you liked this article, please press clap. You can also find the source code on my github page.