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

# # Credit Corp's Consumer Lending Segment
# 
# Note: All numbers are in the form of $'000 unless otherwise stated
# 
# Let's import some libraries first...

# In[1]:


import pandas
from pandas.plotting import scatter_matrix

from sklearn import datasets
from sklearn import model_selection
from sklearn import linear_model

# models
from sklearn.ensemble import RandomForestRegressor
from sklearn.neighbors import KNeighborsRegressor
from sklearn.preprocessing import StandardScaler, PolynomialFeatures
from sklearn.linear_model import LinearRegression, Ridge
from sklearn.tree import DecisionTreeRegressor
from sklearn.metrics import mean_squared_error, r2_score

import matplotlib.pyplot as plt


# Load the past few years of relevant data.

# In[2]:


dataset = pandas.read_csv("data/ccp-consumer-lending-full-year.csv")
print (dataset)


# Let us create a linear regression model with the whole dataset.

# In[3]:


array = dataset.values
    
X = array[:,3:5]   # data = avg_gross_loan_book, net_lending
Y = array[:,2]     # result = NPAT

model = LinearRegression()
model.fit(X, Y)    # train model

# the model's linear regression coefficients
print("Coefficients: \t%s" % model.coef_)
print("Intercept: \t%s" % model.intercept_)

print("\nThe equation would look like...")
print("p = %sr + %sl + %s" % (model.coef_[0], model.coef_[1], model.intercept_))


# Where
# ```
# p = Net profit before tax (npbt)
# b = Average gross loan book (gross_book_average)
# l = Net lending for the period (net_lending)
# ```

# # FY19 Predictions
# 
# ## Based on management's forecasts
# 
# Assumptions:
# 
# * Gross loan book to end the year at $199.896m (long story on how I got to something so specific)
# 
# * Average gross loan book will be $191.496m
# 
# * Net lending will be $50m, on the upper range of the forecast. Quoting a high number here will actually reduce NPBT.
# 

# In[4]:


gross_book_average = 191496
net_lending = 50000

npbt = model.predict([[gross_book_average, net_lending]])[0]

print("EBIT = $%sm" % (npbt/1000))
print("NPAT = $%sm" % (npbt/1000 * 0.7))


# This sits inside the $17 - 19m range forecast by management, so our model is not crazy bad!

# 
# ## Based on a zero-growth scenario
# 
# The higher the Net Lending completed by the company, the lower the reported Net Profit due to the way the company provisions the expected lossed upfront. So you get a situation where the NPAT is under-reported, unless the company stops growing its loan book. So what happens with NPAT when the loan book stops growing?
# 
# * Assume 17.34% of gross loan book is the required net lending to maintain the loan book.
# * Last 5 years (FY14 - FY18) this figure has been: 14.22%, 17.88%, 16.82%, 13.99%, 17.34%.

# In[5]:


net_lending = gross_book_average * 0.1734

print("\nNet Lending Assumption = %s\n" % net_lending)

npbt_zero_growth = model.predict([[gross_book_average, net_lending]])[0]

print("EBIT: $%sm" % (npbt_zero_growth / 1000))
print("NPAT: $%sm" % (npbt_zero_growth * 0.7 / 1000))

print("\nNPAT buffer: $%sm" % ((npbt_zero_growth - npbt) / 1000))