Logistic Regression Practice Exercise

Chemotherapy for Stage B/C colon cancer

Description These are data from one of the first successful trials of adjuvant chemotherapy for
colon cancer. Levamisole is a low-toxicity compound previously used to treat worm infestations
in animals; 5-FU is a moderately toxic (as these things go) chemotherapy agent. There are two
records per person, one for recurrence and one for death

Attribute Information

id: id
study: 1 for all patients
sex: 1=male
age: in years
obstruct: obstruction of colon by tumour
perfor: perforation of colon
adhere: adherence to nearby organs
nodes: number of lymph nodes with detectable cancer
time: days until event or censoring
status: censoring status
differ: differentiation of tumour (1=well, 2=moderate, 3=poor)
extent: Extent of local spread (1=submucosa, 2=muscle, 3=serosa, 4=contiguous
structures)
surg: time from surgery to registration (0=short, 1=long)
node4: more than 4 positive lymph nodes
etype: event type: 1=recurrence,2=death

Loading Libraries

import pandas as pd
from sklearn.linear_model import LogisticRegression
# importing ploting libraries
import matplotlib.pyplot as plt   
#importing seaborn for statistical plots
import seaborn as sns
#Let us break the X and y dataframes into training set and test set. For this we will use
#Sklearn package's data splitting function which is based on random function
from sklearn.model_selection import train_test_split
import numpy as np
# calculate accuracy measures and confusion matrix
from sklearn import metrics
Question 1: Import the Dataset

df=pd.read_csv('colon.csv').drop('Unnamed: 0',axis=1)

Question 2: Get the Dimensionality of the Dataset.

df.shape

(1858, 15)

Question 3: How many Missing Values are there? Drop all missing values.

df.isnull().sum()
df=df.dropna()

Question 4: Generate the five point summary of the data set.

df.describe().T

count mean std min 25% 50% 75% max

id 1776.0 466.506757 269.321338 1.0 234.75 466.5 700.25 929.0

study 1776.0 1.000000 0.000000 1.0 1.00 1.0 1.00 1.0

sex 1776.0 0.518018 0.499816 0.0 0.00 1.0 1.00 1.0

age 1776.0 59.810811 11.911137 18.0 53.00 61.0 69.00 85.0

obstruct 1776.0 0.192568 0.394427 0.0 0.00 0.0 0.00 1.0

perfor 1776.0 0.030405 0.171748 0.0 0.00 0.0 0.00 1.0

adhere 1776.0 0.144144 0.351335 0.0 0.00 0.0 0.00 1.0

nodes 1776.0 3.663288 3.539129 0.0 1.00 2.0 5.00 33.0

status 1776.0 0.493243 0.500095 0.0 0.00 0.0 1.00 1.0

differ 1776.0 2.061937 0.510833 1.0 2.00 2.0 2.00 3.0

extent 1776.0 2.884009 0.478322 1.0 3.00 3.0 3.00 4.0

surg 1776.0 0.268018 0.443052 0.0 0.00 0.0 1.00 1.0

node4 1776.0 0.264640 0.441265 0.0 0.00 0.0 1.00 1.0

time 1776.0 1542.555180 946.741234 8.0 573.00 1856.0 2331.00 3329.0

etype 1776.0 1.500000 0.500141 1.0 1.00 1.5 2.00 2.0

Question 5: How many levels are there in the Dependent Variable?

df.etype.value_counts()

2 888
1 888
Name: etype, dtype: int64

Question 6: With reference to the previous question, plot the levels of the dependent variable in
a plot of your choice.

sns.countplot(df.etype)
plt.grid()
plt.show()

Question 7: Drop the dependent variable from the Data Set and store it seperately. Then split
your data into train and test data sets. The test data size should be 30% of the total data. Use
random_state=7.

X = df.drop('etype',axis=1)
Y = df.etype
test_size = 0.30
seed = 7
X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=test_size, random_stat

Question 8: Formulate a logistic regression model on the train data.

model = LogisticRegression(max_iter=1000)
model.fit(X_train, y_train)
y_predict = model.predict(X_test)

Question 9: Get the Model Score.

model_score = model.score(X_test, y_test)
print('Accuracy Score is ',model_score)

Accuracy Score is 0.5684803001876173

Question 10: Generate the Confusion Matrix and Classification Report. What are your
observations and recommendations?

print('Confusion Matrix','\n',metrics.confusion_matrix(y_test, y_predict),'\n')
print('Classification Report','\n',metrics.classification_report(y_test, y_predict))

Confusion Matrix
[[143 115]
[115 160]]

Classification Report
precision recall f1-score support

1 0.55 0.55 0.55 258

2 0.58 0.58 0.58 275

accuracy 0.57 533

macro avg 0.57 0.57 0.57 533
weighted avg 0.57 0.57 0.57 533

Double-click (or enter) to edit

Try other techniques like CART RF ANN to improve upon accuracy

and play around the model tuning techniques