plot_scatter.py

a = st.beta_expander('About (click to expand)')

a.write("The crossplot analysis allows you to plot two features against one another with the marker sizes representing \
what we might consider as dependant features e.g. total deaths, total deaths per million, total vaccinations.")
a.write("DataFrame masking allows you to explore the data by continent or include all countries.")
a.write("Each feature's central tendency is represented by the dashed line on each axis so you can see where \
countries are positioned in terms of the distribution for each feature. These update when you change features.")
a.write('The heat map is a summary of the correlations between all features')

# DataFrame set up
df_country_unique = df_final[df_final.columns[1:3]].drop_duplicates(subset=['location'], keep='first').set_index(
    'location')
df_analysis = pd.concat([df_country_unique, df_final.groupby('location')[df_final.columns[3:]].max()], axis=1)
df_analysis = df_analysis.reset_index()
df_analysis.rename(columns={'index': 'location'}, inplace=True)

st.subheader('Crossplots')

# a dictionary of string representation keys and feature values (independent variables)
variable_dic = {'GDP per Capita': 'gdp_per_capita',
                'Population': 'population',
                'Stringency Index': 'stringency_index',
                'Population Density': 'population_density',
                'Median Age': 'median_age',
                'Aged 65 or older': 'aged_65_older',
                'Aged 70 or older': 'aged_70_older',
                'Extreme Poverty': 'extreme_poverty',
                'Cardiovascular Death Rate': 'cardiovasc_death_rate',
                'Diabetes Prevalance':'diabetes_prevalence',
                'Female Smokers':'female_smokers',
                'Male Smokers': 'male_smokers',
                'Handwashing Facilities': 'handwashing_facilities',
                'Hospital Beds per Thousand':'hospital_beds_per_thousand',
                'Hospital patients per million':'hosp_patients_per_million',
                'Life Expectancy': 'life_expectancy'}

# a dictionary of string representation keys and feature values (dependent variables)
size_dic = {'Total Deaths': 'total_deaths',
            'Total Deaths per Million': 'total_deaths_per_million',
            'Total Cases per Million': 'total_cases_per_million',
            'People Vaccinated per Hundred': 'people_vaccinated_per_hundred',
            'No sizing': 'no_sizing'}

# set up some columns for the interactive widgets - use a mid point to create some buffer between widgets
c1, mid, c2, = st.beta_columns((3,0.5,3))

# set up the widget options for continents
continents = list(df_analysis.continent.unique())
continents.sort()
continents.insert(0,'Show All Countries')

# set up widget options for x,y, scatter marks and get values from dic
x_metric = c1.selectbox('X axis', list(variable_dic.keys()), 1)
x = variable_dic[x_metric]

y_metric = c1.selectbox('Y axis', list(variable_dic.keys()), 4)
y = variable_dic[y_metric]

size_by = c2.selectbox('Size markers by', list(size_dic.keys()), 0)
size_by = size_dic[size_by]

# if we want to get relative sizes of markers set up plot frame including variable to size
if size_by != 'no_sizing':
    df_plot = df_analysis[[x, y, size_by, 'location', 'continent']]
    rel_sizing = True
else:
    df_plot = df_analysis[[x, y, 'location', 'continent']]
    rel_sizing = False

# select continent
df_continent = c2.selectbox('Show which continent', continents, 0)

# marker size adjustment
if size_by != 'no_sizing':
    marker_size = st.slider('Use this to adjust relative marker size', 1, 1000, step=1,value=150)
else:
    marker_size = st.slider('Use this to adjust marker size', 1, 30, step=1, value=10)

# select which measure of central tendency to show
average_kind = c1.selectbox('Which central tendency', ('Mean', 'Median'), 0)

# if people have selected a specific continent to show then mask DataFrame
if df_continent != 'Show All Countries':
    df_plot = df_plot[df_plot.continent == df_continent]

# function that adjusts the relative marker size using sklearn.preprocessing.normalize
def reshape_for_plot(df, col, marker_size):
    return pd.Series((normalize([df[col]]) * marker_size)[0])

# main function that generates plot
def plot_scatter(df, x, y, size, marker_size, rel_sizing =True, average_kind ='Mean'):

    fig_scatter = go.Figure() # get a graph objects figure
    df_to_plot = df.reset_index(drop=True)
    col = px.colors.qualitative.Plotly * 25 # get 25 colours

    # if we want relative sizing
    if rel_sizing:
        plot_size = reshape_for_plot(df_to_plot, size, marker_size)
        hover_texts = df_to_plot[size]
    else:
        plot_size = [marker_size] * df_to_plot.shape[0]
        hover_texts = [''] * df_to_plot.shape[0]

    # loop through each row and add marker and various bits of meta data
    for i in range(df_to_plot.shape[0]):
        fig_scatter.add_trace(go.Scatter(
            x=np.array(df_to_plot[x][i]),
            y=np.array(df_to_plot[y][i]),
            name=df_to_plot['location'][i],
            hovertext='<b>' + df_to_plot['location'][i] + '</b>' + '<br>' + size.capitalize().replace('_', ' ') + ' : ' + \
                      str(hover_texts[i]),
            hoverinfo="text",
            mode='markers',

            # marker size is adjusted using a reshape function
            marker=dict(size=plot_size[i], opacity=0.5,
                        color=col[i])))

    # add vertical and horizontal lines to represent mean or median
    if average_kind == 'Mean':
        fig_scatter.add_vline(x=df_to_plot[x].mean(), line_width=1, line_dash="dash", line_color="grey")
        fig_scatter.add_hline(y=df_to_plot[y].mean(), line_width=1, line_dash="dash", line_color="grey")
    else:
        fig_scatter.add_vline(x=df_to_plot[x].median(), line_width=1, line_dash="dash", line_color="grey")
        fig_scatter.add_hline(y=df_to_plot[y].median(), line_width=1, line_dash="dash", line_color="grey")

    # use log scale for x and y - makes plot more readable
    fig_scatter.update_xaxes(type="log")
    fig_scatter.update_yaxes(type="log")

    # some formatting of plot - background colours and show legend
    fig_scatter.update_layout(legend={'itemsizing': 'constant'})
    fig_scatter.update_layout(plot_bgcolor='white',
                              width=1000,
                              height=1000,
                              showlegend=True)

    # some formatting of plot - axis font size and tick size
    fig_scatter.update_layout(yaxis=dict(title=y.capitalize().replace('_', ' '), titlefont_size=15, tickfont_size=10),
                              xaxis=dict(title=x.capitalize().replace('_',' '), titlefont_size=15, tickfont_size=10))

    return fig_scatter