7 - Aprendizado de Máquina Estatístico

import math
import os
import random
from pathlib import Path
from collections import defaultdict
from itertools import product
#
import pandas as pd
import numpy as np
#
from sklearn import preprocessing
from sklearn.neighbors import KNeighborsClassifier
from sklearn import metrics
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
#
from xgboost import XGBClassifier
#
from dmba import plotDecisionTree, textDecisionTree
#
import seaborn as sns
import matplotlib.pyplot as plt
from matplotlib.patches import Ellipse

O diretório DATA contém os arquivos .csv utilizados nos exemplos.

DATA = './'

Se você não mantiver seus dados no mesmo diretório que o código, adapte os nomes dos caminhos.

LOAN200_CSV = DATA + 'loan200.csv'
LOAN3000_CSV = DATA + 'loan3000.csv'
LOAN_DATA_CSV = DATA + 'loan_data.csv.gz'

Defina isso se o notebook travar na parte do XGBoost.

os.environ['KMP_DUPLICATE_LIB_OK'] = 'TRUE'

loan200 = pd.read_csv(LOAN200_CSV)
 
predictors = ['payment_inc_ratio', 'dti']
outcome = 'outcome'

newloan = loan200.loc[0:0, predictors]
X = loan200.loc[1:, predictors]
y = loan200.loc[1:, outcome]

knn = KNeighborsClassifier(n_neighbors=20)
knn.fit(X, y)
knn.predict(newloan)
print(knn.predict_proba(newloan))

[[0.45 0.55]]

nbrs = knn.kneighbors(newloan)
maxDistance = np.max(nbrs[0][0])

fig, ax = plt.subplots(figsize=(4, 4))
sns.scatterplot(x='payment_inc_ratio', y='dti', style='outcome', 
                hue='outcome', data=loan200, alpha=0.3, ax=ax)
sns.scatterplot(x='payment_inc_ratio', y='dti', style='outcome', 
                hue='outcome', 
                data=pd.concat([loan200.loc[0:0, :], loan200.loc[nbrs[1][0] + 1,:]]), 
                ax=ax, legend=False)
ellipse = Ellipse(xy=newloan.values[0], 
                  width=2 * maxDistance, height=2 * maxDistance,
                  edgecolor='black', fc='None', lw=1)
ax.add_patch(ellipse)
ax.set_xlim(3, 16)
ax.set_ylim(15, 30)

plt.tight_layout()
plt.show()

loan_data = pd.read_csv(LOAN_DATA_CSV)
loan_data = loan_data.drop(columns=['Unnamed: 0', 'status'])
loan_data['outcome'] = pd.Categorical(
    loan_data['outcome'], 
    categories=['paid off', 'default'], 
    ordered=True)
#
predictors = [
    'payment_inc_ratio','dti',
    'revol_bal','revol_util']
outcome = 'outcome'
#
newloan = loan_data.loc[0:0, predictors]
print(newloan)
X = loan_data.loc[1:, predictors]
y = loan_data.loc[1:, outcome]
#
knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(X, y)
#
nbrs = knn.kneighbors(newloan)
print(X.iloc[nbrs[1][0], :])
#
newloan = loan_data.loc[0:0, predictors]
X = loan_data.loc[1:, predictors]
y = loan_data.loc[1:, outcome]
#
scaler = preprocessing.StandardScaler()
scaler.fit(X * 1.0)
#
X_std = scaler.transform(X * 1.0)
newloan_std = scaler.transform(newloan * 1.0)
#
knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(X_std, y)
#
nbrs = knn.kneighbors(newloan_std)
print(X.iloc[nbrs[1][0], :])

   payment_inc_ratio  dti  revol_bal  revol_util
0             2.3932  1.0       1687         9.4
       payment_inc_ratio   dti  revol_bal  revol_util
35536            1.47212  1.46       1686        10.0
33651            3.38178  6.37       1688         8.4
25863            2.36303  1.39       1691         3.5
42953            1.28160  7.14       1684         3.9
43599            4.12244  8.98       1684         7.2
       payment_inc_ratio   dti  revol_bal  revol_util
2080             2.61091  1.03       1218         9.7
1438             2.34343  0.51        278         9.9
30215            2.71200  1.34       1075         8.5
28542            2.39760  0.74       2917         7.4
44737            2.34309  1.37        488         7.2

loan_data = pd.read_csv(LOAN_DATA_CSV)
loan_data = loan_data.drop(columns=['Unnamed: 0', 'status'])
loan_data['outcome'] = pd.Categorical(
    loan_data['outcome'], 
    categories=['paid off', 'default'], 
    ordered=True)

predictors = ['dti', 'revol_bal', 'revol_util', 'open_acc', 
              'delinq_2yrs_zero', 'pub_rec_zero']
outcome = 'outcome'

X = loan_data[predictors]
y = loan_data[outcome]

knn = KNeighborsClassifier(n_neighbors=20)
knn.fit(X, y)
plt.scatter(
    range(len(X)),
    [bs + random.gauss(0, 0.015) for bs in knn.predict_proba(X)[:,0]], 
    alpha=0.1, marker='.')
knn.predict_proba(X)[:, 0]

loan_data['borrower_score'] = knn.predict_proba(X)[:, 0]
print(loan_data['borrower_score'].describe())

count    45342.000000
mean         0.501091
std          0.128728
min          0.000000
25%          0.400000
50%          0.500000
75%          0.600000
max          0.950000
Name: borrower_score, dtype: float64