ccerv1

import pandas as pd
import requests, zipfile, io

# requests the data from the USDA url
response = requests.get('https://apps.fas.usda.gov/psdonline/downloads/psd_coffee_csv.zip')

# the response is a zipfile, which we unzip in memory rather than saving to the disk
zf = zipfile.ZipFile(io.BytesIO(response.content))
usda = pd.read_csv(zf.open('psd_coffee.csv'))

ccerv1

def country_query(country, year_type='Market_Year', attribute=''):
    df = usda[usda['Country_Name'] == country]
    if attribute:
      df = df[df['Attribute_Description'] == attribute]
    return df.pivot(
      index='Attribute_Description', 
      columns=year_type, 
      values='Value'
    )

ccerv1

def attribute_query(attribute, year_type='Market_Year', countries=[]):
    df = usda[usda['Attribute_Description'] == attribute]
    if countries:
      df = df[df['Country_Name'].isin(countries)]
    return df.pivot(
      index='Country_Name', 
      columns=year_type, 
      values='Value'
    )

ccerv1

%matplotlib inline
import matplotlib.pyplot as plt

# Some constant assumptions we'll use in our analysis
YEARS_IN_SERIES = 25
MAJOR_PRODUCER_THRESHOLD = 5000  # thousands of bags
MIDSIZE_PRODUCER_THRESHOLD = 500 # thousands of bags

ccerv1

def is_producing_above(country_data, threshold_value, num_years_back=5):
    recent_production = country_data.iloc[0,-num_years_back:]
    production_peak = recent_production.mean()
    return production_peak > threshold_value

list_of_all_countries = usda['Country_Name'].unique()
major_producers = []
midsize_producers = []
minor_producers = []

ccerv1

minor_producers_volumes = attribute_query('Production', countries=minor_producers).sum()
midsize_producers_volumes = attribute_query('Production', countries=midsize_producers).sum()
other_producers_volumes = pd.DataFrame(
    data=[minor_producers_volumes, midsize_producers_volumes], 
    index=['Minor producers', 'Mid-size producers']
)
other_producers_volumes.T.plot(kind='area', stacked=True, figsize=(12,8), alpha=.4)

ccerv1

major_producers_volumes = attribute_query('Production', countries=major_producers)
major_producers_volumes = major_producers_volumes.sort_values(2019)
major_producers_volumes.T.plot(kind='area', stacked=True, figsize=(12,8), alpha=.4)

ccerv1

production_table = pd.concat([other_producers_volumes, major_producers_volumes], axis=0)
production_table = production_table.iloc[:,-YEARS_IN_SERIES:]
production_table = production_table.T
production_table

ccerv1

production_table /= 1000     # change to millions instead of thousands of bags

fig, ax = plt.subplots()
production_table.plot(
    kind='area', 
    stacked=True, 
    figsize=(12,8), 
    alpha=0.4, 
    colormap='Blues', 
    legend=False, 

ccerv1

labels = []
y_vals = []
cum_sum = 0
for col_num, country_name in enumerate(production_table.columns):
    labels.append(country_name)
    prod_val = production_table.iloc[-1,col_num]
    y_vals.append((cum_sum * 2 + prod_val)/2)
    cum_sum += prod_val

for i, label in enumerate(labels):