logistic_regression.py

#!/usr/bin/python
import pandas as pd
import statsmodels.api as sm
import math
import matplotlib.pyplot as plt
import numpy as np

loansData = pd.read_csv('loansData_clean.csv')
loansData['IR_TF'] = loansData['Interest.Rate'] < 12

loansData['intercept'] = 1.0

logit = sm.Logit(loansData['IR_TF'], loansData[independentVariablesList])
result = logit.fit()
coeff = result.params
print coeff

# multi variant?
# mx+b // not -b-mx-nz
def interest_rate(FicoScore, AmountRequested, coeff):
    return -(coeff['intercept'] + (coeff['FICO.Score'] * FicoScore) + (coeff['Amount.Requested'] * AmountRequested))

def logistic_function(FicoScore, AmountRequested, coeff):
    return 1 / (1 + math.exp(interest_rate(FicoScore, AmountRequested, coeff)))

print logistic_function(750, 10000, coeff)
# 0.9759220629968894
# 0.024077937003110648

"""
p is > 0.7

Higher FICO Score translates to more chance of getting a loan
with an interest rate less than 12%
"""

# plt.plot(
#     np.linspace(550, 950, num=400),
#     [ logistic_function(coeff, 10000, y) for y in range(550, 950) ]
# )

updated interest rate function