In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
sns.set()
%matplotlib inline
import warnings
warnings.filterwarnings("ignore")
In [2]:
#import data
data = pd.read_csv(r'/Users/shivambadkas/Downloads/liver.csv')
In [3]:
(data.head())
(data.info())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 583 entries, 0 to 582
Data columns (total 11 columns):
 #   Column                      Non-Null Count  Dtype  
---  ------                      --------------  -----  
 0   Age                         583 non-null    int64  
 1   Gender                      583 non-null    object 
 2   Total_Bilirubin             583 non-null    float64
 3   Direct_Bilirubin            583 non-null    float64
 4   Alkaline_Phosphotase        583 non-null    int64  
 5   Alamine_Aminotransferase    583 non-null    int64  
 6   Aspartate_Aminotransferase  583 non-null    int64  
 7   Total_Protiens              583 non-null    float64
 8   Albumin                     583 non-null    float64
 9   Albumin_and_Globulin_Ratio  579 non-null    float64
 10  Dataset                     583 non-null    int64  
dtypes: float64(5), int64(5), object(1)
memory usage: 50.2+ KB
In [4]:
#all features numeric except gender, classifier is 1 for liver disease and 2 for none
print(data.isnull().sum())

Age                           0
Gender                        0
Total_Bilirubin               0
Direct_Bilirubin              0
Alkaline_Phosphotase          0
Alamine_Aminotransferase      0
Aspartate_Aminotransferase    0
Total_Protiens                0
Albumin                       0
Albumin_and_Globulin_Ratio    4
Dataset                       0
dtype: int64
In [5]:
#use scikit imputation to fill in missing values
from sklearn.impute import SimpleImputer
imp = SimpleImputer(missing_values=np.nan, strategy='mean')
data['Albumin_and_Globulin_Ratio']=imp.fit_transform(data[['Albumin_and_Globulin_Ratio']])
In [6]:
data.isnull().sum()
Out[6]:
Age                           0
Gender                        0
Total_Bilirubin               0
Direct_Bilirubin              0
Alkaline_Phosphotase          0
Alamine_Aminotransferase      0
Aspartate_Aminotransferase    0
Total_Protiens                0
Albumin                       0
Albumin_and_Globulin_Ratio    0
Dataset                       0
dtype: int64
In [7]:
#EDA: make correlation heatmap to see which features are correlated if any
X = data.drop(['Gender','Dataset'], axis=1)
plt.figure(figsize=(16, 6))
y = data['Dataset'] 
X.corr().style.background_gradient(cmap='coolwarm')
Out[7]:
Age Total_Bilirubin Direct_Bilirubin Alkaline_Phosphotase Alamine_Aminotransferase Aspartate_Aminotransferase Total_Protiens Albumin Albumin_and_Globulin_Ratio
Age 1.000000 0.011763 0.007529 0.080425 -0.086883 -0.019910 -0.187461 -0.265924 -0.216089
Total_Bilirubin 0.011763 1.000000 0.874618 0.206669 0.214065 0.237831 -0.008099 -0.222250 -0.206159
Direct_Bilirubin 0.007529 0.874618 1.000000 0.234939 0.233894 0.257544 -0.000139 -0.228531 -0.200004
Alkaline_Phosphotase 0.080425 0.206669 0.234939 1.000000 0.125680 0.167196 -0.028514 -0.165453 -0.233960
Alamine_Aminotransferase -0.086883 0.214065 0.233894 0.125680 1.000000 0.791966 -0.042518 -0.029742 -0.002374
Aspartate_Aminotransferase -0.019910 0.237831 0.257544 0.167196 0.791966 1.000000 -0.025645 -0.085290 -0.070024
Total_Protiens -0.187461 -0.008099 -0.000139 -0.028514 -0.042518 -0.025645 1.000000 0.784053 0.233904
Albumin -0.265924 -0.222250 -0.228531 -0.165453 -0.029742 -0.085290 0.784053 1.000000 0.686322
Albumin_and_Globulin_Ratio -0.216089 -0.206159 -0.200004 -0.233960 -0.002374 -0.070024 0.233904 0.686322 1.000000 
<Figure size 1152x432 with 0 Axes>
In [8]:
sns.countplot(data = data, x = 'Dataset', label='Count')
LD, NLD = data['Dataset'].value_counts()
print('Number of patients diagnosed with liver disease: ',LD)
print('Number of patients not diagnosed with liver disease: ',NLD)

Number of patients diagnosed with liver disease:  416
Number of patients not diagnosed with liver disease:  167
In [9]:
sns.countplot(data=data, x = 'Gender', label='Count')
M, F = data['Gender'].value_counts()
print('Number of patients that are male: ',M)
print('Number of patients that are female: ',F)

Number of patients that are male:  441
Number of patients that are female:  142
In [10]:
#import scikit modules
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier as KNN
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
from sklearn.preprocessing import MinMaxScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import roc_curve, roc_auc_score
In [11]:
#create cross validation sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.3, random_state = 101)
print (X_train.shape)
print (y_train.shape)
print (X_test.shape)
print (y_test.shape)

(408, 9)
(408,)
(175, 9)
(175,)
In [12]:
# Decision tree 
dt = DecisionTreeClassifier(max_depth=4,min_samples_leaf=0.14,random_state=1)
dt.fit(X_train,y_train)
dt_y_pred = dt.predict(X_test)
dt_accuracy = accuracy_score(y_test, dt_y_pred)
print('Test set accuracy of dt: {:.2f}'.format(dt_accuracy)) 

Test set accuracy of dt: 0.69
In [13]:
#Logistic Regression
lr = LogisticRegression(solver='lbfgs', multi_class='ovr', max_iter = 1000)
lr.fit(X_train, y_train)
lr_y_pred = lr.predict(X_test)
lr_accuracy = accuracy_score(y_test, lr_y_pred)
print('Test set accuracy of dt: {:.2f}'.format(lr_accuracy)) 

Test set accuracy of dt: 0.70
In [14]:
#K-Nearest Neighbors
KNN_1 = KNN(n_neighbors = 10)
KNN_1.fit(X_train,y_train)
KNN_y_pred = KNN_1.predict(X_test)
KNN_accuracy = accuracy_score(y_test, KNN_y_pred)
print('Test set accuracy of dt: {:.2f}'.format(KNN_accuracy)) 

Test set accuracy of dt: 0.65
In [16]:
#Random Forest Classifier 
rf = RandomForestClassifier(n_estimators=25, random_state=1)
rf.fit(X_train,y_train)
rf_y_pred = rf.predict(X_test)
rf_accuracy = accuracy_score(y_test, rf_y_pred)
print('Test set accuracy of dt: {:.2f}'.format(rf_accuracy)) 

Test set accuracy of dt: 0.69
In [20]:
#Use random forest to see feature importance
importance = pd.Series(data=rf.feature_importances_,
                        index= X_train.columns)
#sort
importance_sorted = importance.sort_values()

importance_sorted.plot(kind='barh', color='lightgreen')
plt.title('Features Importances')
plt.show()
In [19]:
#create tuple of tests to automate. Observe that the pattern follows a define -> fit -> predict -> test accuracy pipeline
classifiers = [('Decision Tree',dt),('Logistic Regression',lr),('KNN',KNN_1), ('Random Forest', rf)]

#for loop to iterate through 

for clf_name, clf in classifiers:
    clf.fit(X_train,y_train)
    y_pred = clf.predict(X_test)
    print('{:s} : {:.3f}'.format(clf_name, accuracy_score(y_test, y_pred)))

Decision Tree : 0.686
Logistic Regression : 0.697
KNN : 0.651
Random Forest : 0.691
In [17]:
#Model Evaluation: ROC Curves; trade-off between recall and specificity, basically 0s get labelled as 1s when 
#you capture more 1s, ROC curve plots sensitivity on y-axis and specificity on y-axis. Need to sort probability 
#of each record being a 1 from most to least likely, then compute cumulative specificity and recall 
#first find probability of a record being 1 
y_score1 = dt.predict_proba(X_test)[:,1]
y_score2 = lr.predict_proba(X_test)[:,1]
y_score3 = KNN_1.predict_proba(X_test)[:,1]
y_score4 = rf.predict_proba(X_test)[:,1]

#Find AUC score, an ineffective classifier will have a value of 0.5. Larger AUC value means better classifier
print('ROC DecisionTree: ', roc_auc_score(y_test, y_score1))
print('ROC Logistic Regression: ', roc_auc_score(y_test, y_score2))
print('ROC KNN: ', roc_auc_score(y_test, y_score3))
print('ROC Random Forest: ', roc_auc_score(y_test, y_score4))

ROC DecisionTree:  0.6856419987349779
ROC Logistic Regression:  0.7365591397849462
ROC KNN:  0.6859582542694497
ROC Random Forest:  0.7283364958886781
In [22]:
results = pd.DataFrame({'Model': ['Decision Tree','Logistic Regression', 'KNN','Random Forest Classifier'],  'Roc_Score': [roc_auc_score(y_test, y_score1), roc_auc_score(y_test, y_score2),roc_auc_score(y_test, y_score3), roc_auc_score(y_test, y_score4)],
                       'Accuracy_Score': [dt_accuracy, lr_accuracy,KNN_accuracy, rf_accuracy]})
df_results = results.sort_values(by='Roc_Score', ascending=False)
df_results
Out[22]:
Model Roc_Score Accuracy_Score
1 Logistic Regression 0.736559 0.697143
3 Random Forest Classifier 0.728336 0.691429
2 KNN 0.685958 0.651429
0 Decision Tree 0.685642 0.685714
In [ ]: