kaggle_kingdom_classifer

Identify the evolutionary origins (Kingdom) of these organisms. https://www.kaggle.com/c/iith-aml-hackathon-2020/overview

Attempt_Hackathon.ipynb

contains feature engineering and different trained models and their classification reports.

Models

Extra Tree Classifier
Random Forest Classifier
XG_BOOST Classifier
Logistic Regression Classifier
SVM Classifier
Neural Network

Result

Best result on validation set was obtained using SVM classifier

training parameters

{
    'kernel':['rbf'],
     'tol': [0.00001,0.000001,0.000001],
     'C': [1,10,100],
     'max_iter': [3000],
     'decision_function_shape': ['ovo'],
     'gamma':['auto'],
     'class_weight': [ {
         0:5,
         1:0.25,
         2:0.25,
         3:0.25,
         4:0.25,
         5:0.5,
         6:0.5,
         7:5,
         8:5,
         9:5} ]
     }

classification report

              precision    recall  f1-score   support

         0.0       0.57      0.33      0.42        12
         1.0       0.93      0.95      0.94       104
         2.0       0.89      0.91      0.90       155
         3.0       0.91      0.93      0.92       174
         4.0       0.87      0.88      0.87       129
         5.0       0.82      0.82      0.82        76
         6.0       0.83      0.80      0.81        30
         7.0       0.62      0.57      0.59        14
         8.0       0.75      0.67      0.71         9
         9.0       1.00      0.62      0.77         8

    accuracy                           0.88       711
   macro avg       0.82      0.75      0.78       711
weighted avg       0.87      0.88      0.88       711