MACHINE LEARNING EXERCISE: CLASSIFICATION

CERVICAL CANCER RISK FACTORS

Models

• Logistic Regression
• K-Nearest Neighbors
• Decision tree Classifier

About

• The dataset was collected at 'Hospital Universitario de Caracas' in Caracas, Venezuela. The dataset comprises demographic information, habits, and historic medical records of 858 patients. Several patients decided not to answer some of the questions because of privacy concerns (missing values).

Target Variable

• (bool) Biopsy: target variable

Features:

1. (int) Age
2. (int) Number of sexual partners
3. (int) First sexual intercourse (age)
4. (int) Num of pregnancies
5. (bool) Smokes
6. (bool) Smokes (years)
7. (bool) Smokes (packs/year)
8. (bool) Hormonal Contraceptives
9. (int) Hormonal Contraceptives (years)
10. (bool) IUD
11. (int) IUD (years)
12. (bool) STDs
13. (int) STDs (number)
14. (bool) STDs:condylomatosis
15. (bool) STDs:cervical condylomatosis
16. (bool) STDs:vaginal condylomatosis
17. (bool) STDs:vulvo-perineal condylomatosis
18. (bool) STDs:syphilis
19. (bool) STDs:pelvic inflammatory disease
20. (bool) STDs:genital herpes
21. (bool) STDs:molluscum contagiosum
22. (bool) STDs:AIDS
23. (bool) STDs:HIV
24. (bool) STDs:Hepatitis B
25. (bool) STDs:HPV
26. (int) STDs: Number of diagnosis
27. (int) STDs: Time since first diagnosis
28. (int) STDs: Time since last diagnosis
29. (bool) Dx:Cancer
30. (bool) Dx:CIN
31. (bool) Dx:HPV
32. (bool) Dx
33. (bool) Hinselmann: [can also be] target variable
34. (bool) Schiller: [can also be] target variable
35. (bool) Cytology: [can also be] target variable

Sources:

• https://www.kaggle.com/loveall/cervical-cancer-risk-classification
• https://archive.ics.uci.edu/ml/datasets/Cervical+cancer+%28Risk+Factors%29#

Import Libraries

##### Standard Libraries #####
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
sns.set_style("whitegrid")
sns.set_context("poster")

%matplotlib inline

##### Other Libraries #####

## Classification Algorithms ##
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn import tree

## For building models ##
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler

## For measuring performance ##
from sklearn import metrics
from sklearn.model_selection import cross_val_score

## To visualize decision tree ##
from sklearn.externals.six import StringIO  
from IPython.display import Image  
from sklearn.tree import export_graphviz
import pydotplus

Load the Dataset

### Load the data
df = pd.read_csv("kag_risk_factors_cervical_cancer_cleaned.csv")
print("Size of dataset:", df.shape)
df.head()

Size of dataset: (411, 34)

	Age	Number of sexual partners	First sexual intercourse	Num of pregnancies	Smokes	Smokes (years)	Smokes (packs/year)	Hormonal Contraceptives	Hormonal Contraceptives (years)	IUD	...	STDs:HPV	STDs: Number of diagnosis	Dx:Cancer	Dx:CIN	Dx:HPV	Dx	Hinselmann	Schiller	Citology	Biopsy
0	18.0	4.0	15.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	15.0	1.0	14.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
2	34.0	1.0	17.0	1.0	0.0	0.0	0.0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
3	42.0	3.0	23.0	2.0	0.0	0.0	0.0	0.0	0.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
4	44.0	3.0	26.0	4.0	0.0	0.0	0.0	1.0	2.0	0.0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

5 rows × 34 columns

### List of columns with its data type
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 411 entries, 0 to 410
Data columns (total 34 columns):
Age                                   411 non-null float64
Number of sexual partners             411 non-null float64
First sexual intercourse              411 non-null float64
Num of pregnancies                    411 non-null float64
Smokes                                411 non-null float64
Smokes (years)                        411 non-null float64
Smokes (packs/year)                   411 non-null float64
Hormonal Contraceptives               411 non-null float64
Hormonal Contraceptives (years)       411 non-null float64
IUD                                   411 non-null float64
IUD (years)                           411 non-null float64
STDs                                  411 non-null float64
STDs (number)                         411 non-null float64
STDs:condylomatosis                   411 non-null float64
STDs:cervical condylomatosis          411 non-null float64
STDs:vaginal condylomatosis           411 non-null float64
STDs:vulvo-perineal condylomatosis    411 non-null float64
STDs:syphilis                         411 non-null float64
STDs:pelvic inflammatory disease      411 non-null float64
STDs:genital herpes                   411 non-null float64
STDs:molluscum contagiosum            411 non-null float64
STDs:AIDS                             411 non-null float64
STDs:HIV                              411 non-null float64
STDs:Hepatitis B                      411 non-null float64
STDs:HPV                              411 non-null float64
STDs: Number of diagnosis             411 non-null float64
Dx:Cancer                             411 non-null float64
Dx:CIN                                411 non-null float64
Dx:HPV                                411 non-null float64
Dx                                    411 non-null float64
Hinselmann                            411 non-null float64
Schiller                              411 non-null float64
Citology                              411 non-null float64
Biopsy                                411 non-null float64
dtypes: float64(34)
memory usage: 109.2 KB

Explore the Dataset

### Get the summary of statistics of the data 
df.describe()

	Age	Number of sexual partners	First sexual intercourse	Num of pregnancies	Smokes	Smokes (years)	Smokes (packs/year)	Hormonal Contraceptives	Hormonal Contraceptives (years)	IUD	...	STDs:HPV	STDs: Number of diagnosis	Dx:Cancer	Dx:CIN	Dx:HPV	Dx	Hinselmann	Schiller	Citology	Biopsy
count	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	...	411.0	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000	411.000000
mean	25.883212	2.155718	17.411192	2.014599	0.048662	0.351097	0.095423	0.666667	1.635693	0.029197	...	0.0	0.026764	0.014599	0.007299	0.014599	0.017032	0.060827	0.116788	0.043796	0.133820
std	7.362462	1.075258	2.745954	1.117393	0.215422	2.434709	0.852453	0.471979	2.848869	0.168564	...	0.0	0.161590	0.120085	0.085227	0.120085	0.129547	0.239304	0.321559	0.204889	0.340874
min	14.000000	1.000000	11.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.0	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	20.000000	1.000000	15.000000	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.0	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
50%	25.000000	2.000000	17.000000	2.000000	0.000000	0.000000	0.000000	1.000000	0.500000	0.000000	...	0.0	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
75%	30.000000	3.000000	18.000000	3.000000	0.000000	0.000000	0.000000	1.000000	2.000000	0.000000	...	0.0	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
max	52.000000	10.000000	29.000000	8.000000	1.000000	34.000000	15.000000	1.000000	20.000000	1.000000	...	0.0	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000

8 rows × 34 columns

### Determine the distribution of values
df["Biopsy"].value_counts()

0.0    356
1.0     55
Name: Biopsy, dtype: int64

As seen on the value counts above, the data is not balanced. Because of this, the models will more likely predict 0.0 more accurately than 1.0.

### Get correlation of features
df.corr()["Biopsy"].sort_values(ascending=False)

Biopsy                                1.000000
Schiller                              0.925147
Hinselmann                            0.647471
Citology                              0.544482
STDs                                  0.441213
STDs: Number of diagnosis             0.421900
STDs (number)                         0.412472
STDs:condylomatosis                   0.334890
STDs:vulvo-perineal condylomatosis    0.334890
Dx                                    0.334890
IUD                                   0.313868
Dx:HPV                                0.309665
Dx:Cancer                             0.309665
Smokes (years)                        0.290817
STDs:HIV                              0.282336
Smokes (packs/year)                   0.257391
IUD (years)                           0.253451
Smokes                                0.243252
Hormonal Contraceptives (years)       0.232390
Dx:CIN                                0.218160
Num of pregnancies                    0.161349
Age                                   0.147161
Number of sexual partners             0.129332
STDs:genital herpes                   0.125647
Hormonal Contraceptives              -0.010107
First sexual intercourse             -0.048507
STDs:cervical condylomatosis               NaN
STDs:vaginal condylomatosis                NaN
STDs:syphilis                              NaN
STDs:pelvic inflammatory disease           NaN
STDs:molluscum contagiosum                 NaN
STDs:AIDS                                  NaN
STDs:Hepatitis B                           NaN
STDs:HPV                                   NaN
Name: Biopsy, dtype: float64

The features Schiller, Hinselmann, Citology and STDS may more likely to be the top predictors for this dataset because of its high correlation with the target variable Biopsy.

Meanwhile, the features with NaN correlation with Biopsy are the columns with only one distinct value, as shown below.

### List columns with only 1 distinct value
print([col for col in df.columns if df[col].value_counts().shape[0] == 1])

['STDs:cervical condylomatosis', 'STDs:vaginal condylomatosis', 'STDs:syphilis', 'STDs:pelvic inflammatory disease', 'STDs:molluscum contagiosum', 'STDs:AIDS', 'STDs:Hepatitis B', 'STDs:HPV']

Prepare the Data for Modelling

First, separate the predictor columns and the target variable. Also, drop the columns with NaN correlation with the target variable Biopsy since it will not contribute on the improvement of the models.

### Separate the predictor columns and the target variable
#### Predictors
X = df.drop(["STDs:cervical condylomatosis", "STDs:vaginal condylomatosis", "STDs:syphilis",
              "STDs:pelvic inflammatory disease", "STDs:molluscum contagiosum", "STDs:AIDS",
              "STDs:Hepatitis B", "STDs:HPV", "Biopsy"], axis=1)

#### Target
y = df["Biopsy"]

print("Shape of X:", X.shape, "\nShape of y:", y.shape)

Shape of X: (411, 25) 
Shape of y: (411,)

Train-Test Split

### Split the dataset into training and test sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=12)

### Check if properly split
print("Shape of X_train:", X_train.shape, "\tShape of X_test:", X_test.shape, 
      "\nShape of y_train:", y_train.shape, "\tShape of y_test:", y_test.shape)

Shape of X_train: (308, 25) 	Shape of X_test: (103, 25) 
Shape of y_train: (308,) 	Shape of y_test: (103,)

Scaling the data using MinMax Scaler

This is performed to make the magnitude of the data more uniform and consistent.

### Instantiate the MinMax Scaler
minmax = MinMaxScaler()

### Fit the scaler to the training set
minmax.fit(X_train)

### Transform the training set
X_train_scaled = minmax.transform(X_train)

### Transform the test set
X_test_scaled = minmax.transform(X_test)

### For cross validation - whole dataset

#### Fit the scaler
minmax.fit(X)

#### Transform the training set
X_cv_scaled = minmax.transform(X)

### Change to Pandas dataframe for easier viewing and manipulation of the data
X_train_sdf = pd.DataFrame(X_train_scaled, index=X_train.index, columns=X_train.columns)
X_test_sdf = pd.DataFrame(X_test_scaled, index=X_test.index, columns=X_test.columns)
X_cv_sdf = pd.DataFrame(X_cv_scaled, index=X.index, columns=X.columns)

Build the Models

### Create a method to plot the confusion matrix for easier viewing
def confmatrix(predicted, title):
    cm = metrics.confusion_matrix(y_test, predicted)

    df_cm = pd.DataFrame(cm, index=[0,1], columns=[0,1])

    fig = plt.figure(figsize= (10,7))
    cmap = sns.diverging_palette(220, 10, as_cmap=True)
    heatmap = sns.heatmap(df_cm,annot=True, fmt="d", cmap=cmap)
    heatmap.yaxis.set_ticklabels(heatmap.yaxis.get_ticklabels(), rotation=0, ha='right', fontsize=16)
    heatmap.xaxis.set_ticklabels(heatmap.xaxis.get_ticklabels(), rotation=0, ha='right', fontsize=16)
    plt.title(title)
    plt.ylabel("True label")
    plt.xlabel("Predicted label")

Logistic Regression

Build/Train the Logistic Regression model

### Instantiate the Algorithm 
logreg = LogisticRegression()

### Train/Fit the model
logreg.fit(X_train_sdf, y_train)

LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False)

Identify feature importance based on the Logistic Regression model

### Check the Trained Model Coefficients
coef = pd.DataFrame(X_train_sdf.columns, columns=["Features"])
coef["Coef"] = logreg.coef_.reshape(-1,1)
coef["| Coef |"] = np.abs(coef["Coef"])
coef.sort_values(by="| Coef |", ascending=False)

	Features	Coef	| Coef |
23	Schiller	3.899563	3.899563
24	Citology	1.624006	1.624006
22	Hinselmann	1.380271	1.380271
2	First sexual intercourse	-0.930836	0.930836
21	Dx	0.885809	0.885809
19	Dx:CIN	0.805565	0.805565
7	Hormonal Contraceptives	-0.574310	0.574310
9	IUD	0.552896	0.552896
8	Hormonal Contraceptives (years)	0.431498	0.431498
1	Number of sexual partners	0.422697	0.422697
11	STDs	0.323547	0.323547
17	STDs: Number of diagnosis	0.287534	0.287534
10	IUD (years)	0.267958	0.267958
4	Smokes	0.261781	0.261781
12	STDs (number)	0.252504	0.252504
0	Age	-0.230098	0.230098
18	Dx:Cancer	0.189293	0.189293
20	Dx:HPV	0.189293	0.189293
14	STDs:vulvo-perineal condylomatosis	0.181461	0.181461
13	STDs:condylomatosis	0.181461	0.181461
5	Smokes (years)	0.178782	0.178782
6	Smokes (packs/year)	0.157158	0.157158
16	STDs:HIV	0.142086	0.142086
3	Num of pregnancies	-0.141888	0.141888
15	STDs:genital herpes	0.000000	0.000000

Based on the magnitude of the coefficients, Schiller, Citology, Hinselmann and First sexual intercourse are the top predictors for this model.

Notice that the predictor First sexual intercourse has the lowest correlation with Biopsy. This just shows that correlation should not always be the main basis on identifying top predictors, especially in classification.

Validate the Logistic Regression model

### Make Predictions
logreg_pred = logreg.predict(X_test_sdf)

### Get the metrics to test performance
logreg_score = metrics.accuracy_score(y_test,logreg_pred) * 100
logreg_recall = metrics.recall_score(y_test, logreg_pred) * 100

### Print Classification report
print("Classification report for classifier %s:\n%s\n"
      % (logreg, metrics.classification_report(y_test, logreg_pred)))

print("Accuracy Score:", logreg_score)

Classification report for classifier LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False):
              precision    recall  f1-score   support

         0.0       0.96      1.00      0.98        89
         1.0       1.00      0.71      0.83        14

    accuracy                           0.96       103
   macro avg       0.98      0.86      0.91       103
weighted avg       0.96      0.96      0.96       103


Accuracy Score: 96.11650485436894

This model has a high accuracy = 98% and a decent recall = 71%.

This confusion matrix shows that the model predicts 0.0 more accurately as expected.

### Plot the confusion matrix
confmatrix(logreg_pred,"Logistic Regression - Cancer - Confusion Matrix")

Cross-validation of the Logistic Regression model

### Perform cross-validation then get the mean
logreg_cv = np.mean(cross_val_score(logreg, X_cv_sdf, y, cv=10)) * 100
print("Cross-Validation Scorev (10 folds):", logreg_cv)

Cross-Validation Scorev (10 folds): 98.29761904761905

K-Nearest Neighbor

Build/Train the KNN Model

### Finding the best K

best_k = 0
best_score = 0

for k in range(3,200,2):
    ### Instantiate the model
    knn = KNeighborsClassifier(n_neighbors=k)

    ### Fit the model to the training set
    knn.fit(X_train_sdf,y_train)
    
    ### Predict on the Test Set
    knn_pred = knn.predict(X_test_sdf)

    ### Get accuracy
    score = metrics.accuracy_score(y_test,knn_pred) * 100
    
    if score >= best_score and k >= best_k:
        best_score = score
        best_k = k

### Print the best score and k
print("---Best results---\nK:", best_k, "\nScore:", best_score)

---Best results---
K: 3 
Score: 96.11650485436894

### Build final model using the best K 
### Set the value of K
k = best_k

### Instantiate the model
knn = KNeighborsClassifier(n_neighbors=k)

### Fit the model to the training set
knn.fit(X_train_sdf,y_train)

### Predict on the Test Set
knn_pred = knn.predict(X_test_sdf)

Validate the KNN model

### Predict on the Test Set
knn_pred = knn.predict(X_test_sdf)

### Get the metrics to test performance
knn_score = metrics.accuracy_score(y_test, knn_pred) * 100
knn_recall = metrics.recall_score(y_test, knn_pred) * 100

### Print Classification report
print("Classification report for classifier %s:\n%s\n"
      % (knn, metrics.classification_report(y_test, knn_pred)))

print("Accuracy Score:", knn_score)

Classification report for classifier KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
                     metric_params=None, n_jobs=None, n_neighbors=3, p=2,
                     weights='uniform'):
              precision    recall  f1-score   support

         0.0       0.96      1.00      0.98        89
         1.0       1.00      0.71      0.83        14

    accuracy                           0.96       103
   macro avg       0.98      0.86      0.91       103
weighted avg       0.96      0.96      0.96       103


Accuracy Score: 96.11650485436894

This model also has a high accuracy = 96% and a decent recall = 71%.

This confusion matrix shows that the model predicts 0.0 more accurately as expected.

### Plot the confusion matrix for easier viewing
confmatrix(knn_pred,"KNN - Cancer - Confusion Matrix")

Cross-validation of the KNN model

### Perform cross-validation then get the mean
knn_cv = np.mean(cross_val_score(knn, X_cv_sdf, y, cv=10)) * 100
print("Cross-validation score (10 folds):", knn_cv)

Cross-validation score (10 folds): 98.05952380952381

Decision Tree Classifier

Build/Train the Decision Tree model

### Instantiate the Algorithm
dtree = tree.DecisionTreeClassifier() #criterion="gini", min_samples_split=2, min_samples_leaf=1, random_state=12)

### Train the model
dtree.fit(X_train_sdf,y_train)

DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort=False,
                       random_state=None, splitter='best')

Validate the Decision Tree model

### Make predictions
dtree_pred = dtree.predict(X_test_sdf)

### Get the metrics to test performance
dtree_score = metrics.accuracy_score(y_test, dtree_pred) * 100
dtree_recall = metrics.recall_score(y_test, dtree_pred) * 100

### Print Classification report
print("Classification report for classifier %s:\n%s\n"
      % (dtree, metrics.classification_report(y_test, dtree_pred)))

print("Accuracy Score:", dtree_score)

Classification report for classifier DecisionTreeClassifier(class_weight=None, criterion='gini', max_depth=None,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort=False,
                       random_state=None, splitter='best'):
              precision    recall  f1-score   support

         0.0       0.97      1.00      0.98        89
         1.0       1.00      0.79      0.88        14

    accuracy                           0.97       103
   macro avg       0.98      0.89      0.93       103
weighted avg       0.97      0.97      0.97       103


Accuracy Score: 97.0873786407767

The 97% accuracy score of this model is also high. This model also has highest recall of 78% compared to the other models tried in this notebook.

Shown below is the confusion matrix.

### Plot the confusion matrix for easier viewing
confmatrix(dtree_pred,"DTree - Cancer - Confusion Matrix")

Visualize Decision Tree model

### Visualize Decision Tree
feature_col = X_train.columns
class_col = pd.unique(y_train)
class_col = np.array(class_col)
class_col = str(class_col).replace(" ", "")

dot_data = StringIO()
export_graphviz(dtree, out_file=dot_data,  
                filled=True, rounded=True,
                special_characters=True,
               feature_names=feature_col,
               class_names=class_col)
graph = pydotplus.graph_from_dot_data(dot_data.getvalue())  
Image(graph.create_png())

pd.DataFrame(dtree.feature_importances_, index = X_train.columns, 
             columns=["Importance"]).sort_values(by="Importance", ascending=False)

	Importance
Schiller	0.916531
Dx:CIN	0.027822
Citology	0.027616
Age	0.022508
Number of sexual partners	0.005522
Smokes	0.000000
STDs:genital herpes	0.000000
First sexual intercourse	0.000000
Hinselmann	0.000000
Dx	0.000000
Dx:HPV	0.000000
Num of pregnancies	0.000000
Dx:Cancer	0.000000
STDs: Number of diagnosis	0.000000
STDs:HIV	0.000000
STDs:vulvo-perineal condylomatosis	0.000000
Smokes (years)	0.000000
STDs:condylomatosis	0.000000
STDs	0.000000
IUD (years)	0.000000
IUD	0.000000
Hormonal Contraceptives (years)	0.000000
Hormonal Contraceptives	0.000000
Smokes (packs/year)	0.000000
STDs (number)	0.000000

The Decision Tree model only used five predictors but still got high accuracy and recall.

All of the predictors for this model are same as the top predictors of the Logistic Regression model except for Num of pregnancies and Number of sexual partners.

Cross-validation of the Decision Tree model

### Perform cross-validation
dtree_cv = np.mean(cross_val_score(dtree, X_cv_sdf, y, cv=10)) * 100
print("Cross-validation score (10 folds):", dtree_cv)

Cross-validation score (10 folds): 98.04761904761905

Summary of Results

summary = pd.DataFrame(index=["K-Nearest Neighbor", "Logistic Regression", "Decision Tree Classifier"],
                       columns=["Accuracy Score", "Recall", "Cross-Validation Score"], 
                       data=[[knn_score, knn_recall, knn_cv], [logreg_score, logreg_recall, logreg_cv],
                             [dtree_score, dtree_recall, dtree_cv]])
summary

	Accuracy Score	Recall	Cross-Validation Score
K-Nearest Neighbor	96.116505	71.428571	98.059524
Logistic Regression	96.116505	71.428571	98.297619
Decision Tree Classifier	97.087379	78.571429	98.047619

Out of all the models tried, the Decision Tree Classifier performed the best compared to the others because of its high accuracy and recall.

The Decision Tree Classifier may have a slightly lower cross-validation score, but its high recall score makes it a better fit for this use case because it is more favorable to incorrectly classify someone with a disease to perform more tests rather than missing someone that actually has a disease.

It is highly recommended to gather more data for a more balanced dataset and try more advanced models. Also, it is recommended to try to fine tune these models since the parameters used here are all default.

Special Thanks

• FTW Foundation