🥕 Instacart vs Me: A Data War
How does my purchasing behavior on specific days of the week, match with the rest of Instacart's users?
How It All Started
After reading Giorgia Lupi & Stefanie Posavec's -- Dear Data, I was inspired to track data on something that was within the confines of my home to understand and analyze real-world observations through data visualization, as an individual class project taught by professor Nick Cawthon.
1 Analyze
The subject matter I chose was to analyze the grocery consumption pattern in my house so I could use that data to make more conscious purchasing and consumption decisions basis of the insights.
Some of the activities I performed before I started to confine my data collection parameters were:
∙ Analyze the items that are currently in the refrigerator that are frequently bought, and disposed of -- Items like bread, eggs, and milk were standard and most frequently bought, but items like fruits and other dairy products were going stale more often and disposed of.
∙ Plan the timeline to start collecting data -- I decided to collect data for the duration of two weeks as that gives me ample time to analyze how often I orders groceries and track repeat orders.
∙ Deciding on which grocery stores to focus on for data collection -- I chose Instacart as my primary source of data collection as it would be easy to maintain since all of the invoicing and transactions happen digitally.
2 Compile
As depicted in the image above, I began to identify each activity I performed on the items I purchased during the period of 28th February to 10th March. These activities included when the item was purchased, consumed, finished, or disposed of due to expiry/going stale. It also included the placement of the items in the kitchen -- outside on the platform, refrigerated, or frozen.
3 Collect
I started to look for public datasets that have information on US grocery purchase data, when I stumbled upon a dataset uploaded by Instacart on Kaggle. This dataset holds data on over 3 million purchase orders from information on -- products, aisles, departments, repeat orders, and timestamp for each order.
4 Combine
I created my own dataset by digitizing the data I captured manually. This was important so that both the datasets I need to compare are in the same format to map together.
I then pulled in the Instacart datasets into Tableau to merge some sheets together. The three sheets I needed from the Instacart dataset were:
1 Products
2 Orders-Products Mapping
3 Orders
2 Orders-Products Mapping
3 Orders
5 Configure
The "Products" sheet held the following data:
∙ Product ID
∙ Product Name
∙ Aisle ID
∙ Department ID
∙ Product ID
∙ Product Name
∙ Aisle ID
∙ Department ID
The "Orders-Products Mapping" sheet held the following data:
∙ Product ID
∙ Order ID
∙ Reordered
∙ Add to Cart Order
∙ Product ID
∙ Order ID
∙ Reordered
∙ Add to Cart Order
The "Orders" sheet held the following data:
∙ Order DOW (Date fo Week)
∙ Order Hour of Day
∙ Order ID
∙ Order Number
∙ User ID
∙ Order DOW (Date fo Week)
∙ Order Hour of Day
∙ Order ID
∙ Order Number
∙ User ID
This is the process I conducted to reach my merged dataset:
Step 1
First, I linked "Products" with "Orders-Products Mapping" with Product ID.
First, I linked "Products" with "Orders-Products Mapping" with Product ID.
Step 2
I then linked "Orders-Products Mapping" and "Orders" with Order ID to get the Order DOW (Date fo Week) into the merged dataset I finally ended up with a dataset that held over 900K rows.
I then linked "Orders-Products Mapping" and "Orders" with Order ID to get the Order DOW (Date fo Week) into the merged dataset I finally ended up with a dataset that held over 900K rows.
Step 3
To match this data with my purchasing info, I filtered this dataset to only show the products I purchased. This brought me down to around 728K results.
To match this data with my purchasing info, I filtered this dataset to only show the products I purchased. This brought me down to around 728K results.
6 Visualize & Observe
This bar chart is a visual representation of the number of Instacart purchases - per day of the week - per product on my shopping list. As I only bought groceries on a Sunday, Monday, and Thursday within the span of two weeks, I compared the rest of Instacart's purchases only for those days.
Here were my findings --
1 I bought Bananas and Bread on all three days of my purchase. But as you notice on this chart, even though I bought most of my products on a Thursday, It is seen that Instacart users bought these two products more often on Sundays and Mondays.
2 Another unusual observation was that people were buying bananas at an unusually high quantity as that of bread or milk (which is categorized as an essential item, and are not likely to buy more bread or milk on Sundays, Mondays, and Thursdays.
3 Both Instacart users and myself buy Strawberries way too much (and often). I bought strawberries twice (Monday and Thursday) in the span of two weeks.
4 The strawberry consumption is a lot higher than other fresh produce.
5 Instacart Users buy more milk on Mondays as opposed to me buying them on Sundays or Thursdays.
🥕
Thanks for reading, here's a carrot.
Professor: Nick Cawthon