Project Summary

Here’s a video of me walking through the entire project in less than 5 minutes:

Project Links

View the Jupyter Notebook

Review my SQL code on GitHub

Business Problem & Key Questions

While superhero movies have grossed a lot over the past decade, this company (confidential) has failed to make a successful (i.e., well-received and profitable) superhero movie.

This time around it conducted research to ensure that the characters are popular and will put on a great show. That way everyone goes to see the movies and people buy merchandise.

I have been hired to analyze their data to answer the following question:

Based on the characters’ popularity and their attributes, who should go head-to-head in the next film?

Since this was a broad question, I developed a few questions to help guide the analysis. Below are my questions:

• Who are the characters in the 'Genius+ IQ' groups?

• Find the characters who are in the 'Genius+' group by also have high durability or great strength

Data Preparation

To obtain the data, I used copied the company’s code for the table that houses their results.

The code can be found here.

Parts of the table were not easily understood. For example, a character’s speed is labeled on a range of 1 - 7, but what does that mean? This is the case for all ability or power-related attributes.

After researching the scales, I created a power grid table based on Marvel Fandom Database

I used SQL to add columns to the company’s table and populate it with the information from the power grid.

I was ready to begin my answering their main question.

Approach

Here’s the approach I used to determine which characters were in the ‘Genius+ IQ’ group:

• I created used a “CASE-WHEN” function to group and count characters by their alignment.

7 characters were in the “Genius+ IQ” group,

Here’s the approach I used to rank the characters by their intelligence among others in their alignment (i.e., good, bad, or anti-hero):

• I used PARTITION BY alignment to ensure that the ranking is done separately for characters with different alignments, such as "Good," "Bad," or "Anti-hero."

• This query ranks provides rankings specific to each alignment category rather than ranking all the characters together.

Here’s the approach I used to find the characters have a genius level IQ or higher, are in the top 25 of popular characters, and also have high durability or great strength:

• I used the query above to create a CTE to help shorten the code.

• strength >= 5, means the character can lift over 25-75 tons over their head

• durability >= 5, means the character is bulletproof

Output:

Results & Key Takeaways

Even though anti-hero films are becoming more popular, the classic, "Good vs Evil" will always fare well.

Only two characters who are in the top 25 of popularity, have a bulletproof level of durability, are classified as a genius or higher, can lift over 25-75 tons over their head, and align with good or bad: Rogue and Doctor Doom!

Based on these analyses, Rogue and Doctor Doom should go head-to-head in a cinematic feature film.

Any comic book fan knows that this dataset is not accurate. I purposely choose to hide the rows from the initial table so you all would follow along despite it being off.

Project Links

View Jupyter Notebook (Full Project)

Review my SQL code on GitHub

Business Problem & Key Questions

Despite my friends and family saying I am a minimalist, I still am striving to practice better. One way I practice is with my wardrobe, so I created a 'capsule wardrobe.'

Before repurposing, donating items, or adding to my wardrobe, I have a few considerations:

• I want the average of items to be at least 2.5 years.

• I like to keep my current items for at least 3 years.

• I consider getting rid of the oldest items first (if there is no attachment/sentiment related to the item)

To decide if I should add new items, I need to answer the following questions :

1. On average, how long have I had the items in my capsule wardrobe?

2. What items have I owned for 3 years or more?

3. Which are the oldest items that I still frequently use?

Data Preparation

To get the information needed to run my analyses, I hand-wrote a list of items. Luckily, it was laundry day and I recently finished relocating so it was relatively easy to catalog.

This list included the following information for each item:

• name

• description

• category

• purpose

• number of years owned

• location

I then converted the list into a table using SQL

Below are 5 random rows from the created table:

To enhance readability and enable more analyses (if wanted), then created a few lookup tables. Two of them are below:

1. Category Table

   

2. Purpose Table

After creating these new tables, I updated the original table with the new information using the foreign keys. I did these in separate steps for demonstrative purposes.

Update with CategoryLookup:

Update with Purpose Lookup:

Here is a snippet of the updated table:

Approach

After engineering the tables to ensure they had all the information needed, the analysis was straightforward. Below are the queries I ran to answer the key questions:

1. Find the average # of years I’ve owned all the items in my wardrobe.

2. What items have I owned for 3 years or more?

3. What are the oldest items in my capsule wardrobe?

Results & Key Takeaways

I plan on going shopping soon!

I’ve owned 10 out of the 20 items in my capsule closet for over three years.

I can look for items like shirts, outerwear, and accessories since I've had items in the category for more than 2.5 years! I have held my shirts and outwear for a long enough time. I can add a few shirts and maybe a new hoodie to my wardrobe.

The oldest items are my Batman chucks and my white pocketed t-shirt. I’ve owned both items for 5 years. The white t-shirt can transition to a hair-drying towel; I do not plan on getting rid of my Batman chucks (EVER!).

Project Summary

Here’s a video of me walking through the entire project in less than 5 minutes:

Project Links

Click to view the Tableau Dashboard

Business Problem & Key Questions

Smiley’s is a thriving electronics store that has multiple stores nationwide. It has remained profitable, despite other electronic stores (i.e., Fry’s and Circuit City) going out of business. The new National Sales Manager wants to ensure Smiley’s continues to do well. As the data analyst on the sales team, I’ve been tasked with creating a report that shows the February sales.

Smiley’s sales team has 5 standard metrics in each report:

1. Number of Customers

2. Quantity of Products Sold

3. Average Items per Customer

4. Average Spent per Customer

5. Total Sales

In addition to those standard metrics, she is interested in the following key questions:

1. What products make up the majority of national sales?

2. Which states/cities performed the highest based on the metrics?

3. Which states/cities performed the worst?

Data Preparation

To prepare my data, I first had to connect the data to Tableau.

I downloaded the two CSV files I needed:

• Sales_February_2019

• customers2

Then I dragged and dropped the tables into the Tableau Data Model.

I create a relationship between the two tables by specifying the relationship so Tableau knows how they're related.

After renaming the necessary columns, I clicked over to the first sheet and started visualizing the data.

Approach

For the 5 standard metrics, I used the following approach:

• COUNTD for the number of customers

• SUM for the quantity of products sold

• Created a calculated field for $ Spent per Customer

  • Items per Order = SUM([Quantity Ordered]) / COUNTD([OrderID])

• Created a calculated field for $ Spent per Customer

  • $ Spent per Customer = ([Total Sales]) / COUNTD([OrdersID])

• Created a calculated field for Total Sales

  • Total Sales = SUM(([Price Each]) * [Quantity Ordered])

Here’s the approach I used to answer see what products generated the best sales and how much of each was sold:

• Created an Order Date Filter to allow users to explore granularity.

• Created “Product Groups” that combine related members in a field

  • For example, MacBook Pro Laptop and ThinkPad Laptop were grouped as Laptops.

  

• After grouping products, I created a table showing the total sales of each group using SUM of Total Sales.

  

• I included a pie chart with a tooltip that shows the percentage of February’s total sales that each group generated.

• I created a bar chart that lists each product for comparison.

  • Each product’s bar is color-coded to show what group it is related to.

Here’s the approach I used to see which states and cities performed best in February:

• Created a separate dashboard displaying a higher-level view

  

  * Dashboard can also be filtered by clicking on the state of interest on the heat map*

• Within this dashboard, there are two key visualizations:

  • a geographic heat map showing the total sales by State

  • a subheader with navigation buttons for the 5 key metrics to allow exploration by city

  • a table that displays the metrics of interest by city

  * These navigation buttons will allow the new National Sales Manager to have options based on the standard metrics on how she wants to judge the performance of the stores*

Given this was the first pass at what she wanted, I would use this dashboard as a launching pad and see if there were any other specific metrics she would like included or to be the focus of the dashboard.

Results & Key Takeaways

Product Performance

1. Laptops and Phones are the best-selling products. The MacBook Pro, Think Pad Laptop, iPhone, and Google Phone are the top sales-generating products.

Store Performance

As the analyst on this project, I am going to recommend the average spent per customer (the table below) because this measure accounts for population differences, unlike the number of products sold or $ in sales.

2. New York, New York was the best-performing store based on the average spent per customer at $206.49.

3. The Portland store in Maine performed the worst by this metric, only averaging $185.38 per customer.