AV Job-A-thon 2021

Approach:

1. First we observe that the feature Credit_Product has missing values. We use imputation here to fill all the values. Here we mark 'Unknown' for each NaN values.
2. Since after training on the data-set we get to know that the feature Credit_Product has highest feature importance. So we will now try to break the "Unknown" value into "U1" and "U0" as per our target variable Is_Lead.
3. Now we want to predict correct Credit_Product featue from rest of our dataset. So we train a RandomForestClassifier for classification of Credit_Product feature. After training, we will add all the probabilities of Credit_Product in the train data itself and for test data also.
4. Now we use CatBoostClassifier to train the data for the target variable Is_Lead.
5. After that we evaluate its roc-auc score.
6. Now we predict the target variable Is_Lead for the test data and save it to Predictions.csv.

• The Notebook file is Main.ipynb.
• The Prediction file is Predictions.csv.
• The EDA is performed in EDA.ipynb.

---

Link to Competition