EM-DAT_natural_disasters.ipynb

Kernel: Python 3

EM-DAT - Natural disasters

Give Feedback | Bug report

Tags: #em-dat #emdat #opendata #analytics #plotly

Author: Jeremy Ravenel

Last update: 2023-04-12 (Created: 2021-05-28)

In 1988, the Centre for Research on the Epidemiology of Disasters (CRED) launched the Emergency Events Database (EM-DAT). EM-DAT was created with the initial support of the World Health Organisation (WHO) and the Belgian Government.

Idea of improvements :

• Put all the curves of natural disasters in a logarithmic graph

Input

In order to use this script, you need to download the dataset on this link : https://public.emdat.be/

• Create an account and connect

• Go to "Data" tab

• Enter your filters criteria

• Press "Download"

Import librairies

import pandas as pd
import plotly.express as px

Variable

PATH_CSV = "path_to_your_file.csv"

Model

Read the CSV

df = pd.read_csv(PATH_CSV)

Configure the plot

# Types
types_df = df[["Year", "Disaster Type"]]
total_line = types_df[["Year"]].value_counts().reset_index(name="value")
total_line["Disaster Type"] = "All"
types_df = types_df.groupby(["Year", "Disaster Type"]).size().reset_index(name="value")
types_df = types_df.append(total_line).sort_values(by=["Year"])

# Countries
count_by_countries = (
    df[["Year", "ISO", "Country"]]
    .groupby(["Year", "ISO", "Country"])
    .size()
    .reset_index(name="counts")
)

Output

fig = px.choropleth(
    count_by_countries,
    locations="ISO",
    color="counts",
    hover_name="Country",
    animation_frame="Year",
    title="Number of natural disasters per country",
    color_continuous_scale=px.colors.sequential.OrRd,
    range_color=[0, count_by_countries["counts"].max()],
)

fig.update_layout(
    width=850, height=600, autosize=False, template="plotly_white", title_x=0.5
)
fig.show()

common_kwargs = {
    "x": "Year",
    "y": "value",
    "title": "Number of natural disasters per year",
}

line_fig = px.line(types_df[types_df["Disaster Type"] == "All"], **common_kwargs)
lineplt_all = px.line(types_df[types_df["Disaster Type"] == "All"], **common_kwargs)
lineplt_filtered = {
    disaster_type: px.line(
        types_df[types_df["Disaster Type"] == disaster_type], **common_kwargs
    )
    for disaster_type in types_df["Disaster Type"].unique()
    if disaster_type != "All"
}
# Add dropdown
line_fig.update_layout(
    updatemenus=[
        dict(
            buttons=list(
                [
                    dict(
                        label="All disasters",
                        method="restyle",
                        args=[{"y": [data.y for data in lineplt_all.data]}],
                    )
                ]
                + [
                    dict(
                        label=disaster_type,
                        method="restyle",
                        args=[
                            {
                                "y": [data.y for data in lineplt.data],
                            }
                        ],
                    )
                    for disaster_type, lineplt in lineplt_filtered.items()
                ]
            ),
        ),
    ],
    title_x=0.5,
    plot_bgcolor="rgba(0,0,0,0)",
)
line_fig.update_xaxes(gridcolor="grey")
line_fig.update_yaxes(gridcolor="grey")
line_fig.show()