Improve accuracy of model using random forest algorithm

#1

well i am getting an accuracy of around 75% .I have tried everything like figuring out the best no.of estimators and have changed max_depth value and everytime i got accuracy below 75% so how can now i proceed to increase its furthur accuracy in kaggle titanic dataset as i see students are also getting 100% accuracy in this dataset

Should i create more features ??

i have cleaned the data

#2

hey @amankharb ,

Other than number of estimators and max_depth , there are other parameters too on which random forest work , so have a look at them at its documentation page . Try hyperparameter tuning to get the best possible parameter values , you can use grid search or random search to get it done.

on Coding Blocks Platform , There are some of our mentors who have submitted the actual test output file while checking this platform and hence achieved 100% accuracy .
You should never go for 100% , always try to be as high as much as possible but less 100 % . 100% looks good but not considered well in data science.

See there proper method or steps to perform any machine learning task :

  1. Understand Data set and data cleaning
  2. feature extraction and selection
  3. Scaling Data , if required
  4. Choosing Model
    a. checking feature importance and neglecting those feature which are not at all useful.
    b. Hyper Parameter Tuning
  5. Prediction or Inference.

If you perform in this way , in correct format, you can achieve really good results on your data .

I hope this would helped you in your doubt.
Thank You and Happy Learning. :slightly_smiling_face:.