From 5f8235b69d290a3729b66acdf449adb8c1a496aa Mon Sep 17 00:00:00 2001
From: Chandrasekhar Ramakrishnan <cramakri@ethz.ch>
Date: Sun, 15 Mar 2020 23:24:56 +0000
Subject: [PATCH] feat: plot trajectory since the 100th case
---
notebooks/Dashboard.ipynb | 135 +++++++++++++++++-
.../covid_19_dashboard/helper.py | 4 +-
2 files changed, 134 insertions(+), 5 deletions(-)
diff --git a/notebooks/Dashboard.ipynb b/notebooks/Dashboard.ipynb
index 6fe07cb7..24e40e7f 100644
--- a/notebooks/Dashboard.ipynb
+++ b/notebooks/Dashboard.ipynb
@@ -7,6 +7,7 @@
"outputs": [],
"source": [
"import pandas as pd\n",
+ "import numpy as np\n",
"import os\n",
"from IPython.display import display, HTML, Markdown\n",
"import covid_19_dashboard as helper"
@@ -158,7 +159,7 @@
"metadata": {},
"outputs": [],
"source": [
- "confirmed_rate_df = helper.growth_df(rates_frames_map, geodata_df, 'confirmed', countries_over_thresh)\n",
+ "confirmed_rate_df = helper.growth_df(rates_frames_map, geodata_df, 'confirmed', countries_over_thresh, 2)\n",
"latest_confirmed_ser = confirmed_rate_df.set_index(\n",
" ['Country/Region', 'Geo Region', 'Date']).drop(\n",
" ['Longitude', 'Latitude'], axis=1).unstack().iloc[:,-1]\n",
@@ -171,7 +172,8 @@
"metadata": {},
"outputs": [],
"source": [
- "base = alt.Chart(confirmed_rate_df).properties(width=300, height=200, title=\"Countries with 2 or more cases per 100k\")\n",
+ "base = alt.Chart(confirmed_rate_df).properties(\n",
+ " width=300, height=200, title=\"Countries with 2 or more cases per 100k\")\n",
"line = base.mark_line().encode(\n",
" x='Date',\n",
" y='Confirmed/100k',\n",
@@ -179,7 +181,134 @@
" facet=alt.Facet('Geo Region:N', columns=1, sort=alt.SortArray(sort_order), title='Geographic Region'),\n",
" tooltip=[\"Country/Region:N\", \"Date:T\", \"Confirmed/100k:Q\"]\n",
")\n",
- "line"
+ "line\n",
+ "display(line)\n",
+ "display(HTML('''\n",
+ "<p style=\"font-size: smaller\">Data Sources: \n",
+ " <a href=\"https://github.com/CSSEGISandData/COVID-19\">JHU CSSE</a>,\n",
+ " <a href=\"https://data.worldbank.org/indicator/SP.POP.TOTL\">World Bank</a>,\n",
+ " <a href=\"https://worldmap.harvard.edu/data/geonode:country_centroids_az8\">Harvard Worldmap</a>\n",
+ "</p>'''))"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "def country_increase_df(c, df_nominal, growth_in_rate_df):\n",
+ " over_100 = df_nominal[df_nominal['Confirmed'] >= 100]\n",
+ " tdf = (over_100[['Date', 'Confirmed']] - over_100.iloc[0][['Date', 'Confirmed']]).reset_index()\n",
+ " tdfr = growth_in_rate_df[(growth_in_rate_df['Date'] >= over_100.iloc[0]['Date']) &\n",
+ " (growth_in_rate_df['Country/Region'] == c)].reset_index()\n",
+ " tdf['Confirmed/100k'] = tdfr['Confirmed/100k']\n",
+ " tdf['Country/Region'] = c\n",
+ " tdf['Days'] = (tdf['Date'] / np.timedelta64(1, 'D')).astype(int)\n",
+ " return tdf[['Country/Region', 'Days', 'Confirmed', 'Confirmed/100k']]\n",
+ "\n",
+ "\n",
+ "growth_in_rate_df = helper.growth_df(rates_frames_map, geodata_df, 'confirmed', countries_over_thresh, 0)\n",
+ "frame_map = {'confirmed': jhu_frames_map['confirmed'].groupby(level='Country/Region').sum()}\n",
+ "growth_in_value_df = helper.growth_df(frame_map, geodata_df, 'confirmed', countries_over_thresh, 1000)\n",
+ "growth_in_value_df = growth_in_value_df.rename({'Confirmed/100k':'Confirmed'}, axis=1)\n",
+ "increase_df = pd.concat([country_increase_df(c, df_nominal, growth_in_rate_df) for \n",
+ " c, df_nominal in growth_in_value_df.groupby('Country/Region')])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "def facetted_growth_plot(df, variable, sort_order, ref_country, title):\n",
+ " base = alt.Chart(df).properties(\n",
+ " width=250, height=150)\n",
+ " line = base.mark_line().encode(\n",
+ " x='Days',\n",
+ " y=variable,\n",
+ " color='Country/Region',\n",
+ " tooltip=[\"Country/Region:N\", \"Days:Q\", f\"{variable}:Q\"]\n",
+ " )\n",
+ " label_loc = increase_df[increase_df['Country/Region'] == ref_country]['Days'].iloc[-2]\n",
+ " ref = base.mark_line(opacity=0.3).encode(\n",
+ " x='Days',\n",
+ " y=variable,\n",
+ " color=alt.ColorValue('steelblue'),\n",
+ " ).transform_filter(f\"datum['Country/Region'] == '{ref_country}'\")\n",
+ " ref += ref.mark_text().encode(text='Country/Region:N').transform_filter(f\"datum['Days'] == {label_loc}\")\n",
+ " charts = []\n",
+ " # make our small multiples\n",
+ " for country in sort_order:\n",
+ " smallm = line.transform_filter(f\"datum['Country/Region'] == '{country}'\").properties(\n",
+ " title=country)\n",
+ " smallm += ref\n",
+ " charts.append(smallm)\n",
+ "\n",
+ " # group the small multiples into 3 horizontal charts\n",
+ " groups = []\n",
+ " c = None\n",
+ " for i, chart in enumerate(charts):\n",
+ " if not i%3:\n",
+ " if c != None:\n",
+ " groups.append(c)\n",
+ " c = alt.hconcat()\n",
+ " c |= chart\n",
+ " # vertically combine the horizontal charts\n",
+ " chart = alt.vconcat(title=title)\n",
+ " for c in groups:\n",
+ " chart &= c\n",
+ " return chart"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "sort_order = growth_in_value_df.groupby(\n",
+ " 'Country/Region').max().sort_values(\n",
+ " 'Confirmed', ascending=False).index.tolist()\n",
+ "# Exclude China in this plot because its numbers are far greater then everywhere else\n",
+ "sort_order = [o for o in sort_order if o != 'China']\n",
+ "chart = facetted_growth_plot(increase_df[increase_df['Country/Region'] != 'China'], \n",
+ " 'Confirmed',\n",
+ " sort_order,\n",
+ " 'Italy',\n",
+ " \"Growth of cases from case 100, compared to Italy\")\n",
+ "display(chart)\n",
+ "display(HTML('''\n",
+ "<p style=\"font-size: smaller\">Data Sources: \n",
+ " <a href=\"https://github.com/CSSEGISandData/COVID-19\">JHU CSSE</a>,\n",
+ " <a href=\"https://data.worldbank.org/indicator/SP.POP.TOTL\">World Bank</a>,\n",
+ " <a href=\"https://worldmap.harvard.edu/data/geonode:country_centroids_az8\">Harvard Worldmap</a>\n",
+ "</p>\n",
+ "<p style=\"font-size: smaller\">Inspired by <a href=\"https://covid19dashboards.com/growth-analysis/\">Thomas Wiecki</a>'''))"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "sort_order = growth_in_value_df.groupby(\n",
+ " 'Country/Region').max().sort_values(\n",
+ " 'Confirmed', ascending=False).index.tolist()\n",
+ "chart = facetted_growth_plot(increase_df, \n",
+ " 'Confirmed/100k',\n",
+ " sort_order,\n",
+ " 'Italy',\n",
+ " \"Growth of cases/100k from case 100, compared to Italy\")\n",
+ "display(chart)\n",
+ "display(HTML('''\n",
+ "<p style=\"font-size: smaller\">Data Sources: \n",
+ " <a href=\"https://github.com/CSSEGISandData/COVID-19\">JHU CSSE</a>,\n",
+ " <a href=\"https://data.worldbank.org/indicator/SP.POP.TOTL\">World Bank</a>,\n",
+ " <a href=\"https://worldmap.harvard.edu/data/geonode:country_centroids_az8\">Harvard Worldmap</a>\n",
+ "</p>'''))"
]
}
],
diff --git a/src/covid-19/covid_19_dashboard/covid_19_dashboard/helper.py b/src/covid-19/covid_19_dashboard/covid_19_dashboard/helper.py
index 6b62468e..fc71c372 100644
--- a/src/covid-19/covid_19_dashboard/covid_19_dashboard/helper.py
+++ b/src/covid-19/covid_19_dashboard/covid_19_dashboard/helper.py
@@ -113,9 +113,9 @@ def map_of_variable(map_df, variable, title):
return p
-def growth_df(rates_frames_map, geodata_df, name, countries_over_thresh):
+def growth_df(rates_frames_map, geodata_df, name, countries_over_thresh, cutoff):
latest_confirmed_ser = rates_frames_map['confirmed'].iloc[:,-1]
- countries_over_1 = latest_confirmed_ser[latest_confirmed_ser >= 2].reset_index()['Country/Region']
+ countries_over_1 = latest_confirmed_ser[latest_confirmed_ser >= cutoff].reset_index()['Country/Region']
confirmed_rate_df = rates_frames_map['confirmed']
confirmed_rate_df = confirmed_rate_df.loc[
--
GitLab