#This program is from the IBM Decision Optimization Documentation

from pandas import DataFrame, Series

names = {
    139987 : "Guadalupe J. Martinez", 140030 : "Michelle M. Lopez", 140089 : "Terry L. Ridgley", 
    140097 : "Miranda B. Roush", 139068 : "Sandra J. Wynkoop", 139154 : "Roland Guérette", 139158 : "Fabien Mailhot", 
    139169 : "Christian Austerlitz", 139220 : "Steffen Meister", 139261 : "Wolfgang Sanger",
    139416 : "Lee Tsou", 139422 : "Sanaa' Hikmah Hakimi", 139532 : "Miroslav Škaroupka", 
    139549 : "George Blomqvist", 139560 : "Will Henderson", 139577 : "Yuina Ohira", 139580 : "Vlad Alekseeva", 
    139636 : "Cassio Lombardo", 139647 : "Trinity Zelaya Miramontes", 139649 : "Eldar Muravyov", 139665 : "Shu T'an", 
    139667 : "Jameel Abdul-Ghani Gerges", 139696 : "Zeeb Longoria Marrero", 139752 : "Matheus Azevedo Melo", 
    139832 : "Earl B. Wood", 139859 : "Gabrielly Sousa Martins", 139881 : "Franca Palermo"}


data = [(139987, "Pension", 0.13221, "Mortgage", 0.10675), (140030, "Savings", 0.95678, "Pension", 0.84446), (140089, "Savings", 0.95678, "Pension", 0.80233), 
                        (140097, "Pension", 0.13221, "Mortgage", 0.10675), (139068, "Pension", 0.80506, "Savings", 0.28391), (139154, "Pension", 0.13221, "Mortgage", 0.10675), 
                        (139158, "Pension", 0.13221, "Mortgage", 0.10675),(139169, "Pension", 0.13221, "Mortgage", 0.10675), (139220, "Pension", 0.13221, "Mortgage", 0.10675), 
                        (139261, "Pension", 0.13221, "Mortgage", 0.10675), (139416, "Pension", 0.13221, "Mortgage", 0.10675), (139422, "Pension", 0.13221, "Mortgage", 0.10675), 
                        (139532, "Savings", 0.95676, "Mortgage", 0.82269), (139549, "Savings", 0.16428, "Pension", 0.13221), (139560, "Savings", 0.95678, "Pension", 0.86779), 
                        (139577, "Pension", 0.13225, "Mortgage", 0.10675), (139580, "Pension", 0.13221, "Mortgage", 0.10675), (139636, "Pension", 0.13221, "Mortgage", 0.10675), 
                        (139647, "Savings", 0.28934, "Pension", 0.13221), (139649, "Pension", 0.13221, "Mortgage", 0.10675), (139665, "Savings", 0.95675, "Pension", 0.27248), 
                        (139667, "Pension", 0.13221, "Mortgage", 0.10675), (139696, "Savings", 0.16188, "Pension", 0.13221), (139752, "Pension", 0.13221, "Mortgage", 0.10675), 
                        (139832, "Savings", 0.95678, "Pension", 0.83426), (139859, "Savings", 0.95678, "Pension", 0.75925), (139881, "Pension", 0.13221, "Mortgage", 0.10675)]

products = ["Savings", "Mortgage", "Pension"]
product_value = [200, 300, 400]
budget_share = [0.2, 0.5, 0.3]

available_budget = 500
channels =  DataFrame(data=[("gift", 20.0, 0.20), ("newsletter", 15.0, 0.05), ("seminar", 23.0, 0.30)], columns=["name", "cost", "factor"])
channels.head(5)

offers = DataFrame(data=data, index=range(0, len(data)), columns=["customerid", "Product1", "Confidence1", "Product2", "Confidence2"])
offers.insert(0,'name', Series(names[i[0]] for i in data))

CSS = """
body {
    margin: 0;
    font-family: Helvetica;
}
table.dataframe {
    border-collapse: collapse;
    border: none;
}
table.dataframe tr {
    border: none;
}
table.dataframe td, table.dataframe th {
    margin: 0;
    border: 1px solid white;
    padding-left: 0.25em;
    padding-right: 0.25em;
}
table.dataframe th:not(:empty) {
    background-color: #fec;
    text-align: left;
    font-weight: normal;
}
table.dataframe tr:nth-child(2) th:empty {
    border-left: none;
    border-right: 1px dashed #888;
}
table.dataframe td {
    border: 2px solid #ccf;
    background-color: #f4f4ff;
}
table.dataframe thead th:first-child {
    display: none;
}
table.dataframe tbody th {
    display: none;
}
"""

from IPython.core.display import HTML
HTML('<style>{}</style>'.format(CSS))

from IPython.display import display
try: 
    display(offers.drop(columns='customerid').sort_values(by='name')) #Pandas >= 0.17
except:
    display(offers.drop(columns='customerid').sort('name')) #Pandas < 0.17

import sys
try:
    import docplex.mp
except:
    raise Exception('Please install docplex. See https://pypi.org/project/docplex/')

try:
    import cplex
except:
    raise Exception('Please install CPLEX. See https://pypi.org/project/cplex/')

from docplex.mp.model import Model

mdl = Model(name="marketing_campaign")
mdl.round_solution = True  # make sure integer vars are automatically rounded

offersR = range(0, len(offers))
productsR = range(0, len(products))
channelsR = range(0, len(channels))

# names of channels
channel_names = channels['name'].tolist()

# this function is used to coin names for channel variables
def name_chan_var(opch):
    offer, p, ch = opch
    return f"ch_{offer}_{p}_{channel_names[ch]}"

channel_vars = mdl.binary_var_cube(offersR, productsR, channelsR, name=name_chan_var)
total_offers = mdl.integer_var(name="total_offers")
budget_spent = mdl.continuous_var(name="spent")

def name_prod_budget(p):
    return f"product_budget_{products[p]}"

budget_per_product = mdl.continuous_var_list(productsR, name=name_prod_budget)

mdl.print_information()

# Only 1 product is offered to each customer     
mdl.add( mdl.sum(channel_vars[o,p,c] for p in productsR for c in channelsR) <= 1
                   for o in offersR)

# total offers is simply sum of channel vars
mdl.add( total_offers == mdl.sum(channel_vars))

# define per product budgets
for p in productsR:
    mdl.add(budget_per_product[p] == mdl.sum(channel_vars[o, p, c] * channels.at[c, "cost"]
                                             for o in offersR
                                             for c in channelsR))
        
mdl.add( budget_spent == mdl.sum(budget_per_product))

# Balance the offers among products
assert sum(budget_share) == 1  # shares equal 1
for p in productsR:
    mdl.add( mdl.sum(channel_vars[o,p,c] for o in offersR for c in channelsR) 
                       <= budget_share[p] * total_offers )
            
# Do not exceed the budget
mdl.add_constraint( budget_spent  <= available_budget )  

mdl.print_information()

expected_p1 =     mdl.sum( channel_vars[idx,p,idx2] * c.factor * product_value[p]* o.Confidence1  
            for p in productsR
            for idx,o in offers[offers['Product1'] == products[p]].iterrows()  
            for idx2, c in channels.iterrows())

expected_p2 =     mdl.sum( channel_vars[idx,p,idx2] * c.factor * product_value[p]* o.Confidence2 
            for p in productsR
            for idx,o in offers[offers['Product2'] == products[p]].iterrows() 
            for idx2, c in channels.iterrows())


mdl.maximize(expected_p1 + expected_p2)

for p in productsR:
    mdl.add_kpi(budget_per_product[p], budget_per_product[p].name)
    
mdl.add_kpi(expected_p1, "Expected P1 return");
mdl.add_kpi(expected_p2, "Expected P2 return");

s = mdl.solve(log_output=True)
assert s, "No Solution !!!"

# print objectives and kpis
mdl.report()

report = [(channels.at[c, "name"], products[p], names[offers.at[o, "customerid"]]) 
          for c in channelsR 
          for p in productsR
          for o in offersR  if abs(channel_vars[o,p,c].solution_value-1) <= 1e-6]

assert len(report) == round(total_offers.solution_value)

print("Marketing plan has {0} offers costing {1}".format(total_offers.solution_value, budget_spent.solution_value))

report_bd = DataFrame(report, columns=['channel', 'product', 'customer'])
display(report_bd)

display(report_bd[report_bd['channel'] == "seminar"].drop(columns='channel'))