Classification Performance
This notebook covers metrics and evaluation techniques for classification models.
Notebook Contents
This notebook covers classification performance evaluation:
- Confusion Matrix: Understanding true/false positives and negatives
- Accuracy Metrics: Precision, recall, F1-score
- ROC Curves: Receiver Operating Characteristic analysis
- Cross-Validation: Model validation techniques
- Performance Visualization: Plotting and interpreting results
Use the buttons above to download the notebook or open it in your preferred environment.
š Notebook Preview
InĀ [1]:
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis, QuadraticDiscriminantAnalysis
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.metrics import classification_report, accuracy_score, f1_score, precision_score, recall_score, confusion_matrix
InĀ [2]:
df = pd.read_csv('Default.csv', index_col=0)
InĀ [3]:
df
Out[3]:
InĀ [4]:
df.iloc[:,1:]
Out[4]:
InĀ [5]:
df.student = df.student.map(dict(Yes=1, No=0))
InĀ [6]:
#df[df.default.isna()]
df
Out[6]:
InĀ [7]:
X_train, X_test, y_train, y_test = train_test_split(df.iloc[:,1:], df.default, test_size=.2, random_state = 1)
InĀ [8]:
len(X_train)
Out[8]:
InĀ [9]:
len(X_test)
Out[9]:
Explore your dataĀ¶
InĀ [10]:
f, ax = plt.subplots(1, 2, figsize = (8,4), sharey= True)
sns.histplot(df.balance,ax = ax[0])
sns.histplot(df.income, ax= ax[1])
Out[10]:
InĀ [Ā ]:
InĀ [21]:
sns.boxplot(x=df.student, y=df.balance)
Out[21]:
InĀ [19]:
sns.boxplot(x=df.default, y=df.income, showmeans=True)
Out[19]:
InĀ [13]:
sns.scatterplot(x = df.balance, y = df.default )
Out[13]:
InĀ [14]:
sns.scatterplot(x = df.balance, y = df.income, hue = df.default,alpha = .7 )
Out[14]:
InĀ [15]:
X_train.iloc[:,1:]
Out[15]:
InĀ [16]:
lda = QuadraticDiscriminantAnalysis()
lda.fit(X_train.iloc[:,1:],y_train)
y_pred_lda = lda.predict(X_test.iloc[:,1:])
con_mat_lda = confusion_matrix(y_test, y_pred_lda)
print(con_mat_lda)
InĀ [24]:
lda = LinearDiscriminantAnalysis()
lda.fit(X_train.iloc[:,1:],y_train)
Out[24]:
InĀ [25]:
qda = QuadraticDiscriminantAnalysis()
qda.fit(X_train.iloc[:,1:],y_train)
Out[25]:
InĀ [28]:
lda.decision_function
Out[28]:
InĀ [31]:
y_pred = qda.predict(X_test.iloc[:,1:])
print('Accuracy of logistic regression classifier on test set: {:.2f}'.format(qda.score(X_test.iloc[:,1:], y_test)))
InĀ [32]:
print(classification_report(y_test, y_pred))
InĀ [38]:
y_pred = lda.predict(X_test.iloc[:,1:])
print('Accuracy of logistic regression classifier on test set: {:.2f}'.format(lda.score(X_test.iloc[:,1:], y_test)))
InĀ [44]:
lda.predict_proba(
InĀ [46]:
y_pred_new_threshold = pd.Series(lda.predict_proba(X_test.iloc[:,1:])[:,1]>=0.6) #(lda.predict_proba(X_test.iloc[:,1:])[:,1]>=0.3).astype(str))
InĀ [30]:
print(classification_report(y_test, y_pred))
InĀ [47]:
print(classification_report(y_test, y_pred_new_threshold))
InĀ [Ā ]:
con_mat = confusion_matrix(y_test, y_pred)
print(con_mat)
int(3)
InĀ [Ā ]:
print(classification_report(y_test, y_pred))
InĀ [Ā ]:
logistic.coef_
InĀ [Ā ]:
logistic.intercept_
InĀ [Ā ]:
sns.scatterplot(x = df.balance, y = df.income, hue = df.default,alpha = .7 )
InĀ [Ā ]:
logistic.predict_proba(X_test)
InĀ [Ā ]:
y_test.unique()
InĀ [45]:
lda.predict_proba(X_test.iloc[:,1:])[:,1]>=0.6
Out[45]:
InĀ [Ā ]:
InĀ [Ā ]:
y_pred_new_threshold.unique()
InĀ [Ā ]:
y_pred_new_threshold = y_pred_new_threshold.map({'False': 'No', 'True': 'Yes'})
InĀ [Ā ]:
y_pred_new_threshold.unique()
InĀ [Ā ]:
con_mat = confusion_matrix(y_test, y_pred_new_threshold)
print(con_mat)