Cumulus Financial Complaints

This was a dataset provided by Real World Fake Data about complaints that have arisen from a financial company since the end of 2011 to October 2020. My goals for this project were to ascertain what areas could be improved upon to help reduce the number of complaints and glean more insight into what products were generating the largest number of complaints. There were no stated business needs but I assumed that they would want to reduce complaints and make their products that were generating the most compliments easier to use.

I started with my data analysis in Microsoft SQL server and then brought those queries into Power BI to visualize the data. It was in Power BI that I noticed that there was maybe some incorrectly labeled data in the products category. The credit card data stopped after 2016 and a new category called credit card prepaid card then started. The checking and savings account data stopped after 2016 as well and Bank account or service continued after. Since I couldn’t verify this error with anyone, I made a judgement call to lump these together. I used the following SQL code to do just that.

-- Changing the products to reflect the correct names
SELECT case when Product = 'Credit card or prepaid card' THEN 'Credit card' when Product = 'Bank account or service' THEN 'Checking or savings account' ELSE Product END FROM [Financial Consumer Complaints]

UPDATE [Financial Consumer Complaints] SET Product = case when Product = 'Credit card or prepaid card' THEN 'Credit card' when Product = 'Bank account or service' THEN 'Checking or savings account' ELSE Product END;

Now after this was done I was able to go thru and get the following information out.

-- number of complaints per product

SELECT product, COUNT(complaint_id) as number_of_complaints FROM[Financial Consumer Complaints] GROUP BY product ORDER BY number_of_complaints desc;

-- number of complaints by product and subproduct
SELECT product, sub_product, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY product, sub_product HAVING COUNT(complaint_id) > 1000 ORDER BY number_of_complaints desc;

-- complaints by product and issue
SELECT product, issue, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY product, issue ORDER BY number_of_complaints DESC;

-- year with highest complaints

SELECT YEAR(Date_Sumbited) as dates, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY YEAR(Date_Sumbited) ORDER BY number_of_complaints DESC;

-- How forms are submitted
SELECT submitted_via as how_submitted, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY submitted_via ORDER BY number_of_complaints DESC;

-- number of complaints in chronological order

SELECT YEAR(date_sumbited) as years, MONTH(date_sumbited) as months, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY year(Date_Sumbited), MONTH(Date_Sumbited) ORDER BY year(Date_Sumbited), MONTH(Date_Sumbited);

-- number of complaints by product sorted chronologically by year
SELECT product, YEAR(date_sumbited) as years, COUNT(complaint_id) as number_of_complaints FROM [Financial Consumer Complaints] GROUP BY product, year(Date_Sumbited) ORDER BY year(Date_Sumbited)

Not all of these queries were used in the final visualization but it was fun to try and pull out as much data as I could.