Product Engineering Review: Marketing_Campaign_Effectiveness_Preditction_PY_SQL

[chetan-hirapara]

Reviewer 1 comments:

• warnings are suppressed.
• SQL Queries are used or DataFrame creation.
• Do not use external plots as it pulls the data to client.
• SQL Queries are used for TD_PLOT

Reviewer 1 suggestions:

• Avoid suppressing the warnings. it may mask some of the genuine issues.
• notebook has a reference for query which has window aggregate on grouping data. It should be modified and use teradataml window aggregates.
• Notebook uses a barplot from matplotlib. teradataml supports barplot. Please look at options and use inline plot if it works for you.
• Notebook uses SQL Query for group by. use teradataml group by instead.
• Use teradataml DataFrame.plot instead of SQL Query.
• Used SQL query for TrainTestSplit instead of teradataml. Use teradataml analytic function.
• Filtered the data based on SQL. Use teradataml dataframe filtering. If we want to store the filtered data, use to_sql on top of filtered DataFrame.
• used SQL for XGBoost. Convert it to teradataml function.
• used SQL for ClassificationEvaluator. Instead, use XGBoostModel.evaluate. That should automatically does evaluation.
• used SQL for DecisionForest. Instead, use teradataml function.

Reviewer 2 comments:

• in section 3.1 pandas shape and and sort used.
• queries used for cast,sum,count function.
• long sql queries used.

Reviewer 2 suggestions:

• use teradataml functions wherever possible