As you can see, in the most recent step (part 5) we found a way to calculate daily new cases. (Note: after doing this, I went back and updated part 2, which is the tutorial where we create the dataset we’ve been using. If you go back to part 2, there is now code there to add daily new cases to the Now that we have the “new cases” variable, we’re going to visualize it.<\/p>\n Our end goal will be to create a “small multiple chart”<\/a> of the daily new cases for the top few countries. Specifically, we’re going to create a small multiple chart of line charts<\/a>.<\/p>\n If you don’t know, I love small multiple charts.<\/p>\n In my opinion, the small multiple chart is one of the most useful but most underused<\/i> charts in the world of data visualization.<\/p>\n It’s a great technique to learn, practice, and master.<\/p>\n Having said that, you often need to wrangle your data into the right shape in order to use it properly. We already did that data wrangling in previous tutorials in this series, but I still have to reiterate: make sure that you learn Pandas<\/i>.<\/p>\n Ok, having said all of that, let’s get to it and create our small multiple chart.<\/p>\n Before you run any of the code in this tutorial, you’ll need to import the proper packages and get the Here, we’ll import Pandas, Seaborn, and datetime.<\/p>\n We’re going to need Pandas to get our dataframe, and to do a little bit of data wrangling on our data (i.e., create subsets, etc).<\/p>\n And we’ll obviously use Seaborn for data visualization.<\/p>\n Next, let’s get the covid-19 dataset.<\/p>\n I retrieved and saved the covid-19 dataset from Johns Hopkins<\/a> as of April 9, 2020. The data is combined and “wrangled” into the proper shape.<\/p>\n You can download the data with this code:<\/p>\n Alternatively, you can run the code in part 2 of the Python covid-19 series<\/a> to create an up-to-date dataset.<\/p>\n We’re actually going to start by creating a single line chart first.<\/p>\n Typically, before I create a full small multiple chart that has multiple panels, I prefer to create a single chart of the chart type that we’ll use in the small multiple panels.<\/p>\n In this case, our final small multiple chart will have line charts.<\/p>\n So that being the case, I want to make a solo line chart just to get a feel for the data and to work out some of the aesthetics.<\/p>\n Let’s start with a simple line chart.<\/p>\n Here, we’ll create a line chart of new covid-19 cases for the USA.<\/p>\n To do that, we’ll subset the First, we’ll use the Pandas query method to subset the rows of our Pandas dataframe<\/a>. We’ll subset down to the rows where Let’s print it out, just to take a look.<\/p>\n OUT:<\/p>\n As you can see, in this subset, we have covid-19 data, by date, for the USA. This includes confirmed cases, but also “new cases”, which we’ll use in our plot.<\/p>\n Now, let’s plot the data using a Seaborn lineplot:<\/p>\n OUT:<\/p>\n Now, we’ll very quickly clean up the formatting.<\/p>\n Here, we’ll change the orientation of the x axis tick labels to 45 degrees, so they don’t overlap.<\/p>\n We’ll also add a title.<\/p>\n OUT:<\/p>\n There’s more that we could possibly do here, but I think that this chart is pretty damn good for a quick-and-dirty line chart.<\/p>\n Now, we’ll take this and create a small multiple chart.<\/p>\n Specifically, we’re going to create a small multiple line chart of the “top 9” countries that have the most new daily cases (as of April 9, 2020).<\/p>\n To do this, we’ll need to retrieve some information about our dataset, wrangle our data into shape, and then plot.<\/p>\n First, we’ll retrieve the top 9 countries, in terms of daily new cases.<\/p>\n To do this, we’ll use several Pandas methods. <\/p>\n We’ll use query to subset the rows down<\/a> to the data for April 9 (the most recent row in this dataset).<\/p>\n Then we’ll use sort_values to sort the data<\/a> by new cases from high to low.<\/p>\n We’ll use the Pandas iloc method<\/a> to retrieve the top 9 rows ….<\/p>\n And then retrieve the The output is actually a Pandas series, but we can retrieve the values as a list.<\/p>\n Let’s print out the countries in OUT:<\/p>\n Next, we’ll use this list to subset our overall dataframe.<\/p>\n Here, we’re going to subset We’re using the Pandas filter method to retrieve a few specific columns of data<\/a> (we don’t need the rest right now).<\/p>\n Then we’re using query to retrieve rows for the countries in our After that, we’re using groupby and agg to compute the total new cases by country, for every date in the dataset.<\/p>\ncovid_data<\/code> dataset.)<\/p>\nVisualizing “daily new cases” using Seaborn<\/h2>\n
![\"A](\"https:\/\/www.sharpsightlabs.com\/wp-content\/uploads\/2020\/04\/covid19-new-cases_small-multiple-chart_top9.png\")
<\/p>\nPreliminary steps (do this first)<\/h3>\n
covid_data<\/code> dataset.<\/p>\nImport packages<\/h5>\n
\r\n#================\r\n# IMPORT PACKAGES\r\n#================\r\nimport pandas as pd\r\nimport seaborn as sns\r\nimport matplotlib.pyplot as plt\r\nimport datetime\r\n<\/pre>\n
Get covid-19 data<\/h5>\n
\r\n#============\r\n# IMPORT DATA\r\n#============\r\ncovid_data = pd.read_csv('https:\/\/learn.sharpsightlabs.com\/datasets\/covid19\/covid_data_2020-04-09.csv'\r\n ,sep = \";\"\r\n )\r\ncovid_data = covid_data.assign(date = pd.to_datetime(covid_data.date, format='%Y-%m-%d'))\r\ncovid_data = covid_data.fillna(value = {'subregion':''})\r\n<\/pre>\nCreate a line chart of new covid19 cases<\/h3>\n
Create a super simple line chart <\/h4>\n
covid_data<\/code> dataset, and then we’ll plot.<\/p>\nSubset data down to USA<\/h5>\n
country<\/code> is “US<\/code>“.<\/p>\n\r\n#------------\r\n# GET US DATA\r\n#------------\r\ncovid_data_US = (covid_data\r\n .query('country == \"US\"')\r\n )\r\n<\/pre>\n\r\nprint(covid_data_US)\r\n<\/pre>\n
\r\n country subregion date ... dead recovered new_cases\r\n18960 US 2020-01-22 ... 0 0.0 NaN\r\n18961 US 2020-01-23 ... 0 0.0 0.0\r\n18962 US 2020-01-24 ... 0 0.0 1.0\r\n18963 US 2020-01-25 ... 0 0.0 0.0\r\n18964 US 2020-01-26 ... 0 0.0 3.0\r\n ... ... ... ... ... ... ...\r\n19034 US 2020-04-05 ... 9619 17448.0 28219.0\r\n19035 US 2020-04-06 ... 10783 19581.0 29595.0\r\n19036 US 2020-04-07 ... 12722 21763.0 29556.0\r\n19037 US 2020-04-08 ... 14695 23559.0 32829.0\r\n19038 US 2020-04-09 ... 16478 25410.0 32385.0\r\n<\/pre>\n
Plot the data<\/h5>\n
\r\nplt.style.use('bmh')\r\nsns.lineplot(data = covid_data_US\r\n ,x = 'date'\r\n ,y = 'new_cases'\r\n )\r\n<\/pre>\n![\"A](\"https:\/\/www.sharpsightlabs.com\/wp-content\/uploads\/2020\/04\/covid-19_USA_daily-new-cases_line-chart.png\")
<\/p>\nClean up the formatting<\/h5>\n
\r\nsns.lineplot(data = covid_data_US\r\n ,x = 'date'\r\n ,y = 'new_cases'\r\n )\r\nplt.xticks(rotation=45, horizontalalignment='right')\r\nplt.title('Covid-19 daily new cases\\nin the United States', fontsize = 18)\r\n<\/pre>\n![\"A](\"https:\/\/www.sharpsightlabs.com\/wp-content\/uploads\/2020\/04\/covid-19_USA_daily-new-cases_line-chart_v2.png\")
<\/p>\nCreate small multiple chart<\/h3>\n
Get “top 9” countries<\/h5>\n
country<\/code> variable.<\/p>\n\r\ncovid_top9_countries = (covid_data\r\n .query('date >= datetime.date(2020, 4, 9)')\r\n .sort_values('new_cases', ascending = False)\r\n .iloc[0:9]\r\n .country\r\n)\r\n<\/pre>\ncovid_top9_countries<\/code>.<\/p>\n\r\nprint(covid_top9_countries.values)\r\n<\/pre>\n
\r\n['US' 'Spain' 'Germany' 'France' 'United Kingdom' 'Italy' 'Turkey' 'Brazil' 'Iran']\r\n<\/pre>\n
Subset dataframe to top 9 countries<\/h5>\n
covid_data<\/code> down to the top 9 countries we just identified.<\/p>\ncovid_top9_countries<\/code> Series.<\/p>\n\r\n#--------------------------------\r\n# CREATE SUBSET: \r\n# - top 9 countries with the most\r\n# new cases\r\n#--------------------------------\r\ncovid_data_country_sub = (covid_data\r\n .filter(['country','date','new_cases'])\r\n .query(\"country in @covid_top9_countries.values\")\r\n .groupby(['country','date'])\r\n .agg('sum')\r\n .reset_index()\r\n)\r\n<\/pre>\n