Overview
As I study for the Associate Professional in Talent Development (APTD) certification, which includes data and analysis skills in the ATD’s Capability Model, I’m repeatedly reminded to update my data analysis skills as an L&D professional. In the early summer of 2024, I plunged into the Cloud Data Analytics certificate course on Google Cloud Platform (GCP), which offered the foundational knowledge of cloud storage and each stage of the data journey. In my capstone project, using product lines on GCP, including AI and automation tools, I applied cloud data skills to answer business questions, prepared a series of reports, and built a self-service dashboard that provides data in actionable ways, enabling the team to proactively manage their lending portfolio, mitigate risks, and make timely decisions.
Objectives
#1: Conducting an analysis of the fintech dataset, including a preliminary report identifying the total number of loans issued by year.
#2: Building a self-service dashboard where the treasury team can interact with key performance metrics.
Approach
Data Journey: Collect, Process, and Store
With defined business questions brought by the Look Fintech team, I looked into the dataset on BigQuery and created the workflow in a data transformation plan. I collected data from a CSV file into a table on the cloud, making it easy to query with the stored dataset in the cloud data warehouse. I transformed the data by writing SQL codes to prepare new tables. I also organized and filtered columns to process reports that answered business questions.
Data Transformation Process
The screencast video captures the process of integrating tables from datasets on BigQuery.
This 1.5-minute video has no audio.
Approach
Data Journey: Analyze and Activate
By pulling datasets from BigQuery to Looker Enterprise, I visualized the data to help the team quickly glance at their loan health and make the most of the historical data, updated hourly or daily on a user-friendly dashboard to draw insights as needed.
Data Visualization Process
The screencast video captures the process of four data visualizations with datasets from BigQuery on Looker.
This 1.5-minute video has no audio.
Delivery
I pulled the results from the analysis and visualization in an executive summary that includes:
- Report with the total number of loans issued by year
- Self-service dashboard screenshot
- Suggestions
Challenges
- Writing SQL language was a technical challenge during the project. Even though the platform didn’t require in-depth knowledge of SQL, spending extra days on SQL tutorials on W3 Schools to obtain a foundational understanding was helpful when the project called for customizing reports.
- Another challenge was understanding the scope of their lending business and the stakeholders’ ultimate needs, including the team and their customers. The business questions and key metrics guided the selection of which data to query that answers the questions. Given the total amount of the outstanding loans was above their threshold, my suggestions focused on how to smoothen the paying-off process for the customers. The team also needed the frequently updated dashboard instead of the historical reports to monitor customer behaviors. I enabled the dashboard’s cross-filtering and refresh rate by date and hour, which allowed the team to obtain the latest data while saving querying costs on the cloud.
Result
Impact: This project provided valuable insights to inform lending decisions, mitigate risk, and tailor financial offerings. I learned to set relevant business objectives and that answers business questions.