Given the task summary, submission instructions, feature engineering script, and model training script below, generate a python script for the final submission:
- Make sure the submission file contains every entity of interest in the test set. If an entity is missing, add a prediction of 0.
- The feature table might contain multiple rows for each entity of interest. Think about which row to keep and only keep one.
- The feature table for model training might not contain features for the test set. In this case, create a feature table for the test set by copying the feature engineering process from the training set.
- Print high-level status in the script. 
- Follow submission instructions closely, especially for column names.
- The final submission file should be saved under the {output_dir} directory as "submission.csv".
- Please generate the script only but nothing else.

Task summary:
{task_summary}

Submission Instructions:
{submission_instructions}

Dataset info:
{dataset_info}

Data Paths:
{data_paths}

Feature Engineering Script:
{fe_code}

Model Training Script:
{train_code}