0
$\begingroup$

I am currently involved in analyzing a particular dataset called Haberman Survival Dataset. The dataset contains cases from a study that was conducted between 1958 and 1970 at the University of Chicago's Billings Hospital on the survival of patients who had undergone surgery for breast cancer.

Attribute Information:

  1. Age of patient at time of operation (numerical)
  2. Patient's year of operation (year - 1900, numerical)
  3. Number of positive axillary nodes detected (numerical)
  4. Survival status (class attribute), 1 = the patient survived 5 years or longer, 2 = the patient died within 5 year

For this dataset I have first processed/cleaned the dataset using R, and now I am planning to collect a sample from this population and then train the sample data (training set) with some suitable machine learning algorithm and build a suitable model (1-3 are predictor variables, 4 is the response variable), and then again collect the sample (test set) from the population and then apply the model that I have built to test set and predict the outcome.

What machine learning models can be used to achieve this objective?

$\endgroup$
1
$\begingroup$

You can use any classification algorithm. Read up on Logistic Regression and Support Vector Machines, if you have sufficient data, you can try neural networks. The threshold score is of extreme importance here for good precision and recall. So lookup, precision-recall curves and ROC curves which help in determining a reasonable threshold.

$\endgroup$
7
  • $\begingroup$ Thank you Mr.Himanshu for your response, I will read up on the Logistic Regression and Support Vector Machines, since I am new to the Field of Data Science is the way of the approach that i am following is correct for this respective Dataset? $\endgroup$
    – shivanesh
    Oct 13 '17 at 14:18
  • $\begingroup$ I have just mentioned a starting point, you will need to create more features, identify important features, perform feature scaling and most importantly identify the difference between correlation and causation. However what I have mentioned is a decent starting point, there is never a one shot solution for ML problems, you need to build upon your previous models and improve as you go along which will require further study and research. I would highly recommend reading up on the Kaggle Titanic survival problem and its tutorials. It will be quite helpful, in your case. $\endgroup$ Oct 13 '17 at 14:23
  • $\begingroup$ Once Again Thank you Mr.Himanshu for your response creating more features in the sense should i create more variables from this dataset?Please forgive if it is a simple question since i am new to data science just a curiosity to know, can you please suggest some tutorial where i can learn more about feature selection? $\endgroup$
    – shivanesh
    Oct 13 '17 at 14:33
  • $\begingroup$ Yes. Please go through the detailed tutorial of the kaggle problem that I have listed. That will answer most of your problems and will help you in tackling your specific problem $\endgroup$ Oct 13 '17 at 14:37
  • $\begingroup$ Thank you soo much Mr.Himanshu i will go through the Kaggle Titanic Survival Dataset and will approach you if i face any issues $\endgroup$
    – shivanesh
    Oct 14 '17 at 10:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.