Historical Events Analysis

Synopsis:

This is an analysis of major disasters in human history with the following objectives:

Understand the nature of disasters and what share of these disasters were avoidable (man-made).
Explore if the pattern of disasters have changed over time, with our ability to predict natural disasters and prevent diseases.
Estimate the impact of these disasters over the current population trends .

Scope, Variables and Datasets:

Categories of major events considered for the analysis:

Natural disasters
Man-made disasters - wars and genocides
Epidemics

All recorded event of above categories with at least 10,000 casualties was considered.

While the original intention was to analyse the trends over last 2000 years , inconsistent data availability limited the scope to a shorter time frame - from 1500 AD to present.

Pre-Processing:

All data sets were sourced from Wikipedia using readHTMLtable function.
Events were filtered for date > 1500 AD and toll > 10,000.
Location field was cleaned and geo-coded for visualization.
Population over time table was loaded for bench marking.

The big picture.

Even at this level of high level visualization it is obvious that:

Man-made disasters account for most of the casualties across the globe.
Asia seems to be have a higher share of natural disasters than other continents.
North America , South America and Australia have done relatively better over the last 500 years.

click for time series visualization.

The details.

To understand the data better , let us explore the trend over time by disaster type.

We can see that

All three disaster categories show increase in occurrence from mid 19th century - better record keeping might possibly explain this.
There is a significant increase in man-made disasters from late 1800s , after a relatively peaceful period of almost a century.
Most of the casualties caused by epidemics is contributed by two major episodes ( early 20th century Influenza and late 20th century HIV)
Overall, it is indisputably clear that man made disasters have contributed to most of the casualties , followed by epidemics and natural disasters in that order.

Population: Regression modeling.

We should also keep in mind that the sharp spike in the death toll over time from mid-18th century might be misleading , as the graph above considers absolute numbers and ignores the sharp increase in population over the time in consideration.

To make the comparison of the magnitude of the events separated by time, it is important that we include population growth trends for the period as well in this analysis.

This introduces a challenge as the population census for all the event-dates is not available - the further we go back in time , less sparse is the census data.

Hence , I built a model to predict population population trend for event dates, using available sparse census over the last 500 years, using a regression model.

Using this population prediction model , I calculated the growth factor and projected the current day equivalent of the casualties by disaster type over time.

Adjusted casualty count.

With this model we can predict current day population equivalent of casualties for these major events.

	Disaster Type	Death Toll
1	man-made	1920
2	epidemics	372
3	natural	44

It is surprising that the casualty count in today’s equivalent population is more than 2.3 billion, almost a third of the current population.

Casualty trend over time:

Casualty by country:

It is interesting to note that China accounts for most of the casualties, Russia comes up as a distant second followed by Congo.

Casualty count by event:

Let us check which are the key events that contributed to most of these casualties.

	Country	Event Name	Death Toll	Date
1	china	Qing dynasty conquest of the Ming dynasty	331	1639
2	multiple	influenza	301	1919
3	china	Taiping Rebellion	188	1858
4	multiple	World War II	185	1942
5	multiple	European colonization of the Americas	159	1696
6	china	Mao Zedong era 1949–1976	150	1962
7	multiple	World War I	124	1916
8	russia	Soviet crimes 1917–1953	110	1935
9	china	Great Chinese Famine	83	1960
10	rome	Thirty Years’ War	80	1633
11	congo	Crimes during Congo Free State 1885–1908	63	1896
12	china	Northern Chinese Famine of 1876–79	58	1878
13	congo	HIV/AIDS	43	1988
14	france	French Wars of Religion	41	1580
15	multiple	Napoleonic Wars	37	1809
16	india	Great Famine of 1876–78	30	1877
17	russia	Russian famine of 1921	30	1922
18	europe	Holocaust	27	1943
19	russia	Russian Civil War	27	1919
20	china	Chinese Civil War	26	1938

It is absolutely shocking to see such high number of (adjusted)casualties against individual events - many of them have resulted in death of 2-5% of entire global population at the time of occurrence. It might be of interest to note that entire US population is about 5% of global population. This table also explains why China comes out at top as the country that has suffered most due to man made disasters.

Conclusion:

While it is widely acknowledged that wars have contributed to exponential advancement in science and technology, we have no model at our disposal to predict how the world would have looked today, had we not lost a population equivalent to almost 2 Billion people due to wars and man-made disasters.

Though the frequency of man made disasters have increased in late 20th and early 21st century the death toll per event has remained comparatively low.

Looking forward, the sad fact is mankind is armed with weapons of mass destruction capable of inflicting damage exponentially higher than what we have suffered till date. Historical data that we have analysed so far gives no confidence in man-kind’s ability to manage conflicts peacefully.

Caveat:

Population trends and projections are based on global averages , have not considered variations in growth trends across continents.

Reference:

Github Source Code