#!/usr/bin/env python
# coding: utf-8

# <a href="https://colab.research.google.com/gist/giabaocorn20/77feb26db4814bdc165577bf47c91043/nobel_prize_analysis.ipynb" target="_parent"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/></a>

# # Setup and Context

# ### Introduction
# 
# On November 27, 1895, Alfred Nobel signed his last will in Paris. When it was opened after his death, the will caused a lot of controversy, as Nobel had left much of his wealth for the establishment of a prize.
# 
# Alfred Nobel dictates that his entire remaining estate should be used to endow “prizes to those who, during the preceding year, have conferred the greatest benefit to humankind”.
# 
# Every year the Nobel Prize is given to scientists and scholars in the categories chemistry, literature, physics, physiology or medicine, economics, and peace.
# 
# <img src=https://i.imgur.com/36pCx5Q.jpg>
# 
# Let's see what patterns we can find in the data of the past Nobel laureates. What can we learn about the Nobel prize and our world more generally?

# ### Upgrade plotly (only Google Colab Notebook)
# 
# Google Colab may not be running the latest version of plotly. If you're working in Google Colab, uncomment the line below, run the cell, and restart your notebook server.

# In[ ]:


get_ipython().run_line_magic('pip', 'install --upgrade plotly')


# ### Import Statements

# In[ ]:


import pandas as pd
import numpy as np
import plotly.express as px
import seaborn as sns
import matplotlib.pyplot as plt


# In[ ]:


from google.colab import drive
drive.mount('/content/drive')


# ### Notebook Presentation

# In[ ]:


pd.options.display.float_format = '{:,.2f}'.format


# ### Read the Data

# In[ ]:


df_data = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/Nobel+Prize+Analysis+Start/nobel_prize_data.csv')


# Caveats: The exact birth dates for Michael Houghton, Venkatraman Ramakrishnan, and Nadia Murad are unknown. I've substituted them with mid-year estimate of July 2nd.
# 

# # Data Exploration & Cleaning
# 
# Preliminary data exploration.
# * What is the shape of `df_data`? How many rows and columns?
# * What are the column names?
# * In which year was the Nobel prize first awarded?
# * Which year is the latest year included in the dataset?

# In[ ]:


print(df_data.shape)
print(df_data.columns)
print("first year:", df_data.year.min())
print("first year:", df_data.year.max())


# 
# * Are there any duplicate values in the dataset?
# * Are there NaN values in the dataset?
# * Which columns tend to have NaN values?
# * How many NaN values are there per column?
# * Why do these columns have NaN values?  

# ### Check for Duplicates

# In[ ]:


print(f'Any duplicates? {df_data.duplicated().values.any()}')


# In[ ]:


print(f'Any NaN values among the data? {df_data.isna().values.any()}')


# In[ ]:


df_data.isna().sum()


# In[ ]:


col_subset = ['year','category', 'laureate_type',
              'birth_date','full_name', 'organization_name']
df_data.loc[df_data.birth_date.isna()][col_subset]


#  We also see that since the organisation's name is in the full_name column, the organisation_name column contains NaN.
# 

# ### Type Conversions
# 
# * Convert the `birth_date` column to Pandas `Datetime` objects
# * Add a Column called `share_pct` which has the laureates' share as a percentage in the form of a floating-point number.

# #### Convert Year and Birth Date to Datetime

# In[ ]:


df_data.birth_date = pd.to_datetime(df_data.birth_date)


# #### Add a Column with the Prize Share as a Percentage

# In[ ]:


separated_values = df_data.prize_share.str.split('/', expand=True)
numerator = pd.to_numeric(separated_values[0])
denomenator = pd.to_numeric(separated_values[1])
df_data['share_pct'] = numerator / denomenator


# In[ ]:


df_data.info()


# In[ ]:


df_data.head()


# # Plotly Donut Chart: Percentage of Male vs. Female Laureates

# **Challenge**: Create a [donut chart using plotly](https://plotly.com/python/pie-charts/) which shows how many prizes went to men compared to how many prizes went to women. What percentage of all the prizes went to women?

# In[ ]:


biology = df_data.groupby("sex").agg({"prize": pd.Series.count})


# In[ ]:


fig = px.pie(labels=biology.index,
             values=biology["prize"],
             hole=.3,
             names=biology.index)
fig.update_traces(textposition='inside', textfont_size=15, textinfo='percent')
fig.show()


# # Who were the first 3 Women to Win the Nobel Prize?
# 
# * What are the names of the first 3 female Nobel laureates?
# * What did the win the prize for?
# * What do you see in their `birth_country`? Were they part of an organisation?

# In[ ]:


df_data[df_data.sex == "Female"].sort_values("year")[:3]


# Even without looking at the data, you might have already guessed one of the famous names: Marie Curie.

# # Find the Repeat Winners
# 
# **Challenge**: Did some people get a Nobel Prize more than once? If so, who were they?

# In[ ]:


is_winner = df_data.duplicated(subset=['full_name'], keep=False)
multiple_winners = df_data[is_winner]
print(f'There are {multiple_winners.full_name.nunique()}' \
      ' winners who were awarded the prize more than once.')


# In[ ]:


col_subset = ['year', 'category', 'laureate_type', 'full_name']
multiple_winners[col_subset]


# We see that Marie Curie actually got the Nobel prize twice - once in physics and once in chemistry. Linus Carl Pauling got it first in chemistry and later for peace given his work in promoting nuclear disarmament. Also, the International Red Cross was awarded the Peace prize a total of 3 times. The first two times were both during the devastating World Wars.

# # Number of Prizes per Category
# 
# * In how many categories are prizes awarded?
# * Create a plotly bar chart with the number of prizes awarded by category.
# * Use the color scale called `Aggrnyl` to colour the chart, but don't show a color axis.
# * Which category has the most number of prizes awarded?
# * Which category has the fewest number of prizes awarded?

# In[ ]:


df_data.category.nunique()


# In[ ]:


prizes_per_category = df_data.category.value_counts()
v_bar = px.bar(
        x = prizes_per_category.index,
        y = prizes_per_category.values,
        color = prizes_per_category.values,
        color_continuous_scale='Aggrnyl',
        title='Number of Prizes Awarded per Category')

v_bar.update_layout(xaxis_title='Nobel Prize Category',
                    coloraxis_showscale=False,
                    yaxis_title='Number of Prizes')
v_bar.show()


# 
# * When was the first prize in the field of Economics awarded?
# * Who did the prize go to?

# In[ ]:


df_data[df_data.category == 'Economics'].sort_values('year')[:3]


# The first Economics prize was awarded in 1969 and it went to Jan Tinbergen.

# # Male and Female Winners by Category
# 
# Create a [plotly bar chart](https://plotly.com/python/bar-charts/) that shows the split between men and women by category.
# * Hover over the bar chart. How many prizes went to women in Literature compared to Physics?
# 

# In[ ]:


cat_men_women = df_data.groupby(['category', 'sex'],
                               as_index=False).agg({'prize': pd.Series.count})
cat_men_women.sort_values('prize', ascending=False, inplace=True)


# In[ ]:


v_bar_split = px.bar(x = cat_men_women.category,
                     y = cat_men_women.prize,
                     color = cat_men_women.sex,
                     title='Number of Prizes Awarded per Category split by Men and Women')

v_bar_split.update_layout(xaxis_title='Nobel Prize Category',
                          yaxis_title='Number of Prizes')
v_bar_split.show()


# We see that overall the imbalance is pretty large with physics, economics, and chemistry. Women are somewhat more represented in categories of Medicine, Literature and Peace. Splitting bar charts like this is an incredibly powerful way to show a more granular picture.

# # Number of Prizes Awarded Over Time
# 
# **Challenge**: created? Show the trend in awards visually.
# * Count the number of prizes awarded every year.
# * Create a 5 year rolling average of the number of prizes (Hint: see previous lessons analysing Google Trends).
# * Using Matplotlib superimpose the rolling average on a scatter plot.
# * Show a tick mark on the x-axis for every 5 years from 1900 to 2020. (Hint: you'll need to use NumPy).
# 
# 
# 
# * Use the [named colours](https://matplotlib.org/3.1.0/gallery/color/named_colors.html) to draw the data points in `dogerblue` while the rolling average is coloured in `crimson`.
# |
# 
# * Looking at the chart, did the first and second world wars have an impact on the number of prizes being given out?
# * What could be the reason for the trend in the chart?
# 

# In[ ]:


prize_per_year = df_data.groupby("year").agg({'prize': pd.Series.count})
prize_per_year.sort_values('year', inplace=True)


# In[ ]:


prize_per_year


# In[ ]:


moving_average = prize_per_year.rolling(window=5).mean()
moving_average


# In[ ]:


plt.figure(figsize=(16,8), dpi=200)
plt.title('Number of Nobel Prizes Awarded per Year', fontsize=18)
plt.yticks(fontsize=14)
plt.xticks(ticks=np.arange(1900, 2021, step=5),
           fontsize=14,
           rotation=45)

ax = plt.gca() # get current axis
ax.set_xlim(1900, 2020)

ax.scatter(x=prize_per_year.index,
           y=prize_per_year.values,
           c='dodgerblue',
           alpha=0.7,
           s=100,)

ax.plot(prize_per_year.index,
        moving_average.values,
        c='crimson',
        linewidth=3,)

plt.show()


# # Are More Prizes Shared Than Before?
# 
# Investigate if more prizes are shared than before.
# 
# * Calculate the average prize share of the winners on a year by year basis.
# * Calculate the 5 year rolling average of the percentage share.
# * Copy-paste the cell from the chart you created above.
# * Modify the code to add a secondary axis to your Matplotlib chart.
# * Plot the rolling average of the prize share on this chart.
# * See if you can invert the secondary y-axis to make the relationship even more clear.

# In[ ]:


yearly_avg_share = df_data.groupby(by='year').agg({'share_pct': pd.Series.mean})
share_moving_average = yearly_avg_share.rolling(window=5).mean()


# In[ ]:


plt.figure(figsize=(16,8), dpi=200)
plt.title('Number of Nobel Prizes Awarded per Year', fontsize=18)
plt.yticks(fontsize=14)
plt.xticks(ticks=np.arange(1900, 2021, step=5),
           fontsize=14,
           rotation=45)

ax1 = plt.gca()
ax2 = ax1.twinx() # create second y-axis
ax1.set_xlim(1900, 2020)

ax1.scatter(x=prize_per_year.index,
           y=prize_per_year.values,
           c='dodgerblue',
           alpha=0.7,
           s=100,)

ax1.plot(prize_per_year.index,
        moving_average.values,
        c='crimson',
        linewidth=3,)
# Can invert axis
ax2.invert_yaxis()

# Adding prize share plot on second axis
ax2.plot(prize_per_year.index,
        share_moving_average.values,
        c='grey',
        linewidth=3,)

plt.show()


# there is clearly an upward trend in the number of prizes being given out as more and more prizes are shared. Also, more prizes are being awarded from 1969 onwards because of the addition of the economics category. We also see that very few prizes were awarded during the first and second world wars. Note that instead of there being a zero entry for those years, we instead see the effect of the wards as missing blue dots.

# # The Countries with the Most Nobel Prizes

# **Challenge**:
# * Create a Pandas DataFrame called `top20_countries` that has the two columns. The `prize` column should contain the total number of prizes won.
# 
# <img src=https://i.imgur.com/6HM8rfB.png width=350>
# 
# * Is it best to use `birth_country`, `birth_country_current` or `organization_country`?
# * What are some potential problems when using `birth_country` or any of the others? Which column is the least problematic?
# * Then use plotly to create a horizontal bar chart showing the number of prizes won by each country. Here's what you're after:
# 
# <img src=https://i.imgur.com/agcJdRS.png width=750>
# 
# * What is the ranking for the top 20 countries in terms of the number of prizes?

# If you look at the entries in the birth country, you'll see that some countries no longer exist! These include the Soviet Union or Czechoslovakia for example. Hence, using birth_country_current is better, since it has the country name which controls the city where the laureate was born. Now, notice that this does not determine the laureates' nationality since some globetrotting folks gave birth to their future Nobel laureate children while abroad. Also, people's nationalities can change as they emigrate and acquire different citizenship or get married and change citizenship. What this boils down to is that we will have to be clear about the assumptions that we will make in the upcoming analysis.

# In[ ]:


top_countries = df_data.groupby("birth_country_current", as_index = False).agg({"prize":pd.Series.count})
top_countries.sort_values(by='prize', inplace=True)
top20_countries = top_countries[-20:]


# In[ ]:


h_bar = px.bar(x=top20_countries.prize,
               y=top20_countries.birth_country_current,
               orientation='h',
               color=top20_countries.prize,
               color_continuous_scale='Viridis',
               title='Top 20 Countries by Number of Prizes')

h_bar.update_layout(xaxis_title='Number of Prizes',
                    yaxis_title='Country',
                    coloraxis_showscale=False)
h_bar.show()


# In[ ]:





# In[ ]:





# # Use a Choropleth Map to Show the Number of Prizes Won by Country
# 
# * Create this choropleth map using [the plotly documentation](https://plotly.com/python/choropleth-maps/):
# 
# <img src=https://i.imgur.com/s4lqYZH.png>
# 
# * Experiment with [plotly's available colours](https://plotly.com/python/builtin-colorscales/). I quite like the sequential colour `matter` on this map.
# 
# Hint: You'll need to use a 3 letter country code for each country.
# 

# In[ ]:


df_countries = df_data.groupby(['birth_country_current', 'ISO'],
                               as_index=False).agg({'prize': pd.Series.count})
df_countries.sort_values('prize', ascending=False)


# In[ ]:


fig = px.choropleth(df_countries, locations="ISO",
                    color="prize", # lifeExp is a column of gapminder
                    hover_name="birth_country_current", # column to add to hover information
                    color_continuous_scale=px.colors.sequential.matter)
fig.show()


# # In Which Categories are the Different Countries Winning Prizes?
# 
# **Challenge**: See if you can divide up the plotly bar chart you created above to show the which categories made up the total number of prizes. Here's what you're aiming for:
# 
# <img src=https://i.imgur.com/iGaIKCL.png>
# 
# * In which category are Germany and Japan the weakest compared to the United States?
# * In which category does Germany have more prizes than the UK?
# * In which categories does France have more prizes than Germany?
# * Which category makes up most of Australia's nobel prizes?
# * Which category makes up half of the prizes in the Netherlands?
# * Does the United States have more prizes in Economics than all of France? What about in Physics or Medicine?
# 
# 
# The hard part is preparing the data for this chart!
# 
# 
# *Hint*: Take a two-step approach. The first step is grouping the data by country and category. Then you can create a DataFrame that looks something like this:
# 
# <img src=https://i.imgur.com/VKjzKa1.png width=450>
# 

# In[ ]:


cat_prize_per_country = df_data.groupby(["birth_country_current", "category"], as_index = False).agg({"prize":pd.Series.count})
cat_prize_per_country.rename(columns={"prize": "cat_prize"}, inplace=True)
cat_prize_per_country


# Join `cat_prize_per_country` with `top_countries`

# In[ ]:


merged_df = pd.merge(cat_prize_per_country, top20_countries, on='birth_country_current')
# change column names
merged_df.columns = ['birth_country_current', 'category', 'cat_prize', 'total_prize']
merged_df.sort_values(by='total_prize', inplace=True)
merged_df


# In[ ]:


cat_cntry_bar = px.bar(x=merged_df.cat_prize,
                       y=merged_df.birth_country_current,
                       color=merged_df.category,
                       orientation='h',
                       title='Top 20 Countries by Number of Prizes and Category')

cat_cntry_bar.update_layout(xaxis_title='Number of Prizes',
                            yaxis_title='Country')
cat_cntry_bar.show()


# we see is that the US has won an incredible proportion of the prizes in the field of Economics. In comparison, Japan and Germany have won very few or no economics prize at all. Also, the US has more prizes in physics or medicine alone than all of France's prizes combined. On the chart, we also see that Germany won more prizes in physics than the UK and that France has won more prizes in peace and literature than Germany, even though Germany has been awarded a higher total number of prizes than France.
# 
# 

# ### Number of Prizes Won by Each Country Over Time
# 
# * When did the United States eclipse every other country in terms of the number of prizes won?
# * Which country or countries were leading previously?
# * Calculate the cumulative number of prizes won by each country in every year. Again, use the `birth_country_current` of the winner to calculate this.
# * Create a [plotly line chart](https://plotly.com/python/line-charts/) where each country is a coloured line.

# In[ ]:


prize_by_year = df_data.groupby(by=['birth_country_current', 'year'], as_index=False).count()
prize_by_year = prize_by_year.sort_values('year')[['year', 'birth_country_current', 'prize']]
prize_by_year


# In[ ]:


cumulative_prizes = prize_by_year.groupby(by=['birth_country_current',
                                              'year']).sum().groupby(level=[0]).cumsum()
cumulative_prizes.reset_index(inplace=True)
cumulative_prizes


# In[ ]:


l_chart = px.line(cumulative_prizes,
                  x='year',
                  y='prize',
                  color='birth_country_current',
                  hover_name='birth_country_current')

l_chart.update_layout(xaxis_title='Year',
                      yaxis_title='Number of Prizes')

l_chart.show()


# What we see is that the United States really started to take off after the Second World War which decimated Europe. Prior to that, the Nobel prize was pretty much a European affair. Very few laureates were chosen from other parts of the world. This has changed dramatically in the last 40 years or so. There are many more countries represented today than in the early days. Interestingly we also see that the UK and Germany traded places in the 70s and 90s on the total number of prizes won. Sweden being 5th place pretty consistently over many decades is quite interesting too. Perhaps this reflects a little bit of home bias?

# # What are the Top Research Organisations?
# 
# **Challenge**: Create a bar chart showing the organisations affiliated with the Nobel laureates. It should looks something like this:
# 
# <img src=https://i.imgur.com/zZihj2p.png width=600>
# 
# * Which organisations make up the top 20?
# * How many Nobel prize winners are affiliated with the University of Chicago and Harvard University?

# In[ ]:


sorted_organization_data = df_data.groupby("organization_name", as_index= False).agg({"prize": pd.Series.count})
sorted_organization_data.sort_values("prize", ascending = False, inplace = True)
sorted_organization_data.reset_index()


# In[ ]:


sorted_organization_data = sorted_organization_data[:20]
sorted_organization_data


# In[ ]:


h_bar = px.bar(x=sorted_organization_data.prize,
               y=sorted_organization_data.organization_name,
               orientation='h',
               color=sorted_organization_data.prize,
               color_continuous_scale='Viridis',
               title='Top 20 Organizations by Number of Prizes')

h_bar.update_layout(xaxis_title='Number of Prizes',
                    yaxis_title='Organizations',
                    coloraxis_showscale=False)
h_bar.show()


# # Which Cities Make the Most Discoveries?
# 
# Where do major discoveries take place?  
# 
# * Create another plotly bar chart graphing the top 20 organisation cities of the research institutions associated with a Nobel laureate.
# * Where is the number one hotspot for discoveries in the world?
# * Which city in Europe has had the most discoveries?

# In[ ]:


top20_org_cities = df_data.organization_city.value_counts()[:20]
top20_org_cities.sort_values(ascending=True, inplace=True)
city_bar2 = px.bar(x = top20_org_cities.values,
                  y = top20_org_cities.index,
                  orientation='h',
                  color=top20_org_cities.values,
                  color_continuous_scale=px.colors.sequential.Plasma,
                  title='Which Cities Do the Most Research?')

city_bar2.update_layout(xaxis_title='Number of Prizes',
                       yaxis_title='City',
                       coloraxis_showscale=False)
city_bar2.show()


# In[ ]:





# # Where are Nobel Laureates Born? Chart the Laureate Birth Cities
# 
# * Create a plotly bar chart graphing the top 20 birth cities of Nobel laureates.
# * Use a named colour scale called `Plasma` for the chart.
# * What percentage of the United States prizes came from Nobel laureates born in New York?
# * How many Nobel laureates were born in London, Paris and Vienna?
# * Out of the top 5 cities, how many are in the United States?
# 

# In[ ]:


top20_cities = df_data.birth_city.value_counts()[:20]
top20_cities.sort_values(ascending=True, inplace=True)
city_bar = px.bar(x=top20_cities.values,
                  y=top20_cities.index,
                  orientation='h',
                  color=top20_cities.values,
                  color_continuous_scale=px.colors.sequential.Plasma,
                  title='Where were the Nobel Laureates Born?')

city_bar.update_layout(xaxis_title='Number of Prizes',
                       yaxis_title='City of Birth',
                       coloraxis_showscale=False)
city_bar.show()


# A higher population definitely means that there's a higher chance of a Nobel laureate to be born there. New York, Paris, and London are all very populous. However, Vienna and Budapest are not and still produced many prize winners. That said, much of the ground-breaking research does not take place in big population centres, so the list of birth cities is quite different from the list above. Cambridge Massachusets, Stanford, Berkely and Cambridge (UK) are all the places where many discoveries are made, but they are not the birthplaces of laureates.

# # Plotly Sunburst Chart: Combine Country, City, and Organisation
# 
# * Create a DataFrame that groups the number of prizes by organisation.
# * Then use the [plotly documentation to create a sunburst chart](https://plotly.com/python/sunburst-charts/)
# * Click around in your chart, what do you notice about Germany and France?
# 
# 
# Here's what you're aiming for:
# 
# <img src=https://i.imgur.com/cemX4m5.png width=300>
# 
# 

# In[ ]:


country_city_org = df_data.groupby(by=['organization_country',
                                       'organization_city',
                                       'organization_name'], as_index=False).agg({'prize': pd.Series.count})

country_city_org = country_city_org.sort_values('prize', ascending=False)


# In[ ]:


burst = px.sunburst(country_city_org,
                    path=['organization_country', 'organization_city', 'organization_name'],
                    values='prize',
                    title='Where do Discoveries Take Place?',
                   )

burst.update_layout(xaxis_title='Number of Prizes',
                    yaxis_title='City',
                    coloraxis_showscale=False)

burst.show()


# # Patterns in the Laureate Age at the Time of the Award
# 
# How Old Are the Laureates When the Win the Prize?
#  Calculate the age of the laureate in the year of the ceremony and add this as a column called `winning_age` to the `df_data` DataFrame. Hint: you can use [this](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.dt.html) to help you.
# 
# 

# In[ ]:


birth_years = df_data.birth_date.dt.year


# In[ ]:


df_data['winning_age'] = df_data.year - birth_years


# ### Who were the oldest and youngest winners?
# 
# **Challenge**:
# * What are the names of the youngest and oldest Nobel laureate?
# * What did they win the prize for?
# * What is the average age of a winner?
# * 75% of laureates are younger than what age when they receive the prize?
# * Use Seaborn to [create histogram](https://seaborn.pydata.org/generated/seaborn.histplot.html) to visualise the distribution of laureate age at the time of winning. Experiment with the number of `bins` to see how the visualisation changes.

# In[ ]:


display(df_data.nlargest(n=1, columns='winning_age'))
display(df_data.nsmallest(n=1, columns='winning_age'))


# In[ ]:


df_data.winning_age.describe()


# ### Descriptive Statistics for the Laureate Age at Time of Award
# 
# * Calculate the descriptive statistics for the age at the time of the award.
# * Then visualise the distribution in the form of a histogram using [Seaborn's .histplot() function](https://seaborn.pydata.org/generated/seaborn.histplot.html).
# * Experiment with the `bin` size. Try 10, 20, 30, and 50.  

# In[ ]:


plt.figure(figsize=(8, 4), dpi=200)
sns.histplot(data=df_data,
             x=df_data.winning_age,
             bins=30)
plt.xlabel('Age')
plt.title('Distribution of Age on Receipt of Prize')
plt.show()


# ### Age at Time of Award throughout History
# 
# Are Nobel laureates being nominated later in life than before? Have the ages of laureates at the time of the award increased or decreased over time?
# 
# 
# * Use Seaborn to [create a .regplot](https://seaborn.pydata.org/generated/seaborn.regplot.html?highlight=regplot#seaborn.regplot) with a trendline.
# * Set the `lowess` parameter to `True` to show a moving average of the linear fit.
# * According to the best fit line, how old were Nobel laureates in the years 1900-1940 when they were awarded the prize?
# * According to the best fit line, what age would it predict for a Nobel laureate in 2020?
# 

# In[ ]:


plt.figure(figsize=(8,4), dpi=200)
with sns.axes_style("whitegrid"):
    sns.regplot(data=df_data,
                x='year',
                y='winning_age',
                lowess=True,
                scatter_kws = {'alpha': 0.4},
                line_kws={'color': 'black'})

plt.show()


# ### Winning Age Across the Nobel Prize Categories
# 
# How does the age of laureates vary by category?
# 
# * Use Seaborn's [`.boxplot()`](https://seaborn.pydata.org/generated/seaborn.boxplot.html?highlight=boxplot#seaborn.boxplot) to show how the mean, quartiles, max, and minimum values vary across categories. Which category has the longest "whiskers"?
# * In which prize category are the average winners the oldest?
# * In which prize category are the average winners the youngest?

# In[ ]:


plt.figure(figsize=(8,4), dpi=200)
with sns.axes_style("whitegrid"):
    sns.boxplot(data=df_data,
                x='category',
                y='winning_age')

plt.show()


# 
# * Now use Seaborn's [`.lmplot()`](https://seaborn.pydata.org/generated/seaborn.lmplot.html?highlight=lmplot#seaborn.lmplot) and the `row` parameter to create 6 separate charts for each prize category. Again set `lowess` to `True`.
# * What are the winning age trends in each category?
# * Which category has the age trending up and which category has the age trending down?
# * Is this `.lmplot()` telling a different story from the `.boxplot()`?
# * Create another chart with Seaborn. This time use `.lmplot()` to put all 6 categories on the same chart using the `hue` parameter.
# 

# In[ ]:


with sns.axes_style('whitegrid'):
    sns.lmplot(data=df_data,
               x='year',
               y='winning_age',
               row = 'category',
               lowess=True,
               aspect=2,
               scatter_kws = {'alpha': 0.6},
               line_kws = {'color': 'black'},)

plt.show()


# We see that winners in physics, chemistry, and medicine have gotten older over time. The ageing trend is strongest for physics. The average age used to be below 50, but now it's over 70. Economics, the newest category, is much more stable in comparison. The peace prize shows the opposite trend where winners are getting younger! As such, our scatter plots showing the best fit lines over time and our box plot of the entire dataset can tell very different stories!