akashbangalkar / cancer-diagnosis

Machine Learning - Multiclass Classification

Personalized Cancer Diagnosis

Business Problem

Data: Memorial Sloan Kettering Cancer Center (MSKCC)

Download training_variants.zip and training_text.zip from Kaggle.

Context: Source: https://www.kaggle.com/c/msk-redefining-cancer-treatment/discussion/35336#198462

Problem statement: Classify the given genetic variations/mutations based on evidence from text-based clinical literature.

Real-World/Business Objectives and Constraints

• No low-latency requirement.
• Interpretability is important.
• Errors can be very costly.
• Probability of a data-point belonging to each class is needed.

Data Overview

• Source: https://www.kaggle.com/c/msk-redefining-cancer-treatment/data
• We have two data files: one conatins the information about the genetic mutations and the other contains the clinical evidence (text) that human experts/pathologists use to classify the genetic mutations.
• Both these data files are have a common column called ID Data file's information:
  • training_variants (ID , Gene, Variations, Class)
  • training_text (ID, Text)

Preprocessing Text Data

Checking Null Values

Merging both gene variations and text data

Perform Exploratory Data Analysis

Univariate Analysis

• Response Coding with Laplace Smoothing
• Univariate Analysis on Gene Feature
• Featurizing Gene Feature
• Univariate Analysis on Variation Feature
• Featurizing Variation Feature
• Univariate Analysis on Text Feature

Click Here To Check Total Work on Case Study.

Machine Learning Models

Stacking all the three types of features

Model	Log Loss
Naive Bayes	1.3007
KNN Classification	1.0827
Logistic Regression (With Class Balancing)	1.2266
Logistic Regression (Without Class Balancing)	1.2201
Linear Support Vector Machines	1.2283
Random Forest Classifier	1.1755
Stacking Model	1.2749
Maximum Voting Classifier	1.2356