Data Analyst

Technical Skills: Power BI, SQL, AWS, Python

My medium articles about data analysis, business tips and other stuff: https://medium.com/@douglasbittencourt

Professional Experience

STAFF do Brasil • https://www.staffdobrasil.com.br/site/ • (70 employees) - Data Analyst; Valinhos, São Paulo, BR | Power BI, MS SQL Server, AWS (Nov 2023 - Now)

• Lead a data warehouse construction project from scratch to consolidate data from two different ERPs, dealing with data ingestion, security, and performance, reducing data processing time by 25%

• Migrate all the dashboards of the company from GoodData to Power BI, rebuilding from scratch, giving many visual improvements to improve readability, and offering real-time and reliable data for better decision-making speed

• Build and maintain Dashboards to support the whole company

Cogna Educação • https://www.kroton.com.br/ • (36,000 employees) - Data Analyst; Valinhos, São Paulo, BR | Python, Power BI, PostgreSQL, MS Excel, SAP (Sep 2022 - Nov 2023)

• Worked at the financial department, following the whole student’s journey looking for continuous improvement through QA

• Automated the messaging and billing system for students with late payment using Python, avoiding having to send it manually

• Data analysis and building Dashboards using SQL and Power BI, to generate insights for managerial decision making.

Elettromec • https://elettromec.com.br/ • (182 employees) - Junior Market Intelligence Analyst; Valinhos, São Paulo, BR | Python, Power BI, SQL Server, MS Excel (Oct 2021 - Sep 2022)

• Supported the sales team with reports, presentations, and dashboards using Power BI, which enhanced productivity and sales insights.

• Automated routines using Python, helping to reduce operational work time.

• Conducted data extraction, storage, and analysis using MS Excel, SQL, and MS Access. Performed market and competitor studies to identify opportunities.

Education

Associate Degree in Software Analysis and Development (2020 - 2022)

UNIP - Campinas, São Paulo, BR

Graduate Course in Neuromarketing · (2016 - 2017)

INOVA BUSINESS SCHOOL - Campinas, São Paulo, BR

Degree in Business Administration with a concentration in Marketing (2012 - 2016)

ESAMC - Campinas, São Paulo, BR

Projects

Geocoding

I made a script in Python for geocoding using Google Maps API - https://github.com/doougbr/Geocoding-Script

Perfume Market Analysis for Data Portfolio

I made this project to demonstrate knowledge in using the tools throughout the entire data pipeline creation process, up to delivering a final report in the form of a dashboard.

One of my passions is perfumery, so I decided to dive into a database with perfume listings sourced from eBay to analise it and take valuables insights from it

Tools: Python (Pandas, Google Cloud and native libraries), Google Cloud Storage, BigQuery, Power BI, GitHub

Processes: ETL, Data Ingestion Automation, API Consumption, Dashboard Creation

Steps:

Download the database (via API)
Process data using Python and automate the upload to the GCS bucket
Create the dataset and tables in BigQuery
Consume data from BigQuery using Power BI and create a dashboard

Process Breakdown:

First, I installed the Kaggle library

pip install kaggle

To use the API, it’s necessary to create a folder called .kaggle in your home directory and save the JSON with the key there.

The official API documentation shows how to query datasets.

For uploading to GCS, I installed the library:

pip install google-cloud-storage

I then created a simple Python script that transforms the table into a dataframe, saves a Parquet file, and automatically uploads it to a bucket in Google Storage. The choice of Parquet was due to its excellent format with very reduced size and the ability to maintain metadata.

I uploaded another similar dataset, but in CSV, to show the size difference between the files.

Next, I created a dataset and a table with the data in BigQuery, reading from the previously created bucket.

I ran a basic query to validate the table.

After that, I made the connection in Power BI with BigQuery.

I made some transformations in Power Query, such as merging the datasets and separating the location by city, state, and country.

Finally, I created a Power BI dashboard with excellent performance, using only native visuals and following good design practices, such as a clean layout, few color variations, contrast, and consistency.

Lastly, I committed the project files to GitHub for version control.

Insights

Regarding the initial questions, this analysis provided some answers to the questions that were originally asked:

Q: What are the best-selling perfumes, brands, types, and gender categories?

A: The best-selling perfume was CK One. The brand was Calvin Klein, which had more than double the sales of the third place, Davidoff.

In terms of type, Eau de Toilette dominated, with almost 3x more sales than Eau de Parfum, and men’s fragrances outsold women’s.
Q: Which brands have the highest added value?

A: Claude Marsal perfumes have the highest average ticket price, but they are rarely sold.
Q: What was the average amount spent on perfumes during this period?

A: The average amount was $42.61, with men spending slightly more on average than women.

For sales on eBay targeting the U.S. market (which represents the majority of the dataset), the recommendation would be to focus on Versace perfumes for men, which have a very high sales volume and a relatively higher average ticket than Calvin Klein.

Sales of niche brands are much lower than designer brands, which suggests that consumers looking for niche perfumes probably do not use eBay as their primary shopping platform. Among the 10 best-selling brands, only Armaf is considered niche.

A potential research suggestion based on this initial analysis could be to investigate how many sellers are offering each brand and/or perfume, in order to perform a competition analysis.