GSPANN’s advanced analytics team examined the requirement and developed POCs for a model deployment using Google Cloud Platform (GCP) and big data tools. We deployed the model on Airflow and built an automated scoring pipeline. Later, we retrained and fine-tuned the model.
Our team analyzed the result of each campaign by applying various data transformations that include calculations, logic alterations, and more, as per the client’s requirements. We rectified data discrepancies (due to duplicate entries) in the relational database, incremental data received daily, and removed the data of the expired campaigns.
Based on the input data, the model determines the cost of execution of each campaign for the given interval based on the start and end date. We used the Managed File Transfer (MFT) process to transfer the files into GCP buckets.
We handled two types of data – Dimensions (full refresh tables) and Fact Data (incremental tables). Using Hive and BigQuery, data was processed into structured Hive tables that were used to create insight tables as per the business requirement. The client’s marketing team visualize the data using Tableau.