Predict Liver Patient With Random Forest And Logistic Regression

Following are the highlights, hopefully some can capture your attention!

• The dataset ILPD (Indian Liver Patient Dataset) [1] comprises 583 instances with each having 10 features and 1 target variable. The dataset is used to classify if a patient, given a feature vector, has the Liver Disease or not (binary classification).
• The predictive algorithms Random Forest and Logistic Regression are chosen for this task.
• The task is made possible thanks to Python, and especially Scikit-Learn/Pandas libraries. Indeed, I used Anaconda3 [2] for “all-in-one” installation.
• GridSearchCV is used to automatically search for optimal parameters in Random Forest and Logistic Regression.
• I missed ggplot2 in R, but in Python for Data Science, seaborn [3] seems promising. This library is used to make Feature Scatter plot.
• The algorithms’ performance is compared using ROC (Receiver operating characteristic) and AUC (Area Under Curve).
• The quickest way to reproduce this report is to run “python.exe ClassificationTask.py” with “DataPreparation.py” and “Indian Liver Patient Dataset (ILPD).csv” put in the same place.

Please see my article:

• https://thachtranerc.wordpress.com/2016/05/15/predict-liver-disease-with-random-forest-and-logistic-regression/

Software Environment:

• Anaconda3 v4.0.0 64bit (Python v3.5.2)
• PyCharm 2016.1.3
• IPython Notebook

Reference:

1. ILPD (Indian Liver Patient Dataset) Data Set

2. Anaconda3

3. Seaborn: statistical data visualization