Illustration by Isabela Humphrey
Words by Jack Riewe
Buckle in. We’re going shopping. 🛒
This Use Case is Dedicated to Retail
This use case goes out to small business owners in the retail space. In retail, there are quite a few emerging tech trends entering the stores we love. Micro-moments, recommendation engines, and predictive analytics can all be accomplished with machine learning. For this use case we will focus on predictive analytics for profit.
About the Retail Dataset
The dataset we will be using is publicly available data of a global superstore over a 4 year period.
The columns in this dataset include:
- Row ID
- Order ID
- Order Date
- Ship Date
- Ship Mode
- Customer ID
- Customer Name
- Postal Code
- Product ID
- Product Name
- Shipping Cost
- Order Priority
Keep in mind, we have cleaned this data and removed all—if any—corrupted rows.
Our Analytic and Prediction Process
If you don’t know by now, we’re a no-code data science tool. We have taken the traditional time-consuming machine learning process and simplified it into 3 steps.
With this dataset, we can upload into our platform, ask questions, and get an algorithmic output in English.
Comparatively the traditional process of preparing and importing data would look something like this:
The Queries We’ll Ask to Predict Profit
We will structure this use case a little differently. We will simply predict profit from the dataset and explore the top factors.
At the end of this post, we will list some ideas of other insights you can get from this dataset to try yourself.
We’ve been asked about our algorithms a lot recently, so in our use cases we will begin to include the tech specs of predictions.
Exploring the Results
After predicting profit, you can instantly see the top driving factors. Obviously, a big factor related to profit is sales. We will skip that and go into the other factors.
Looking at the graph, you can identify the sweet spot for supplying your customers a discount. When you reach a discount around 25%, you will lose profit and continue to lose profit with the bigger discount you give. Obviously, this prediction is based off of the data in the dataset. With your own data, this is incredible for finding your own discount equilibrium.
Next, we will look at the type of categories related to profit. This is the dataset’s “Sub-Category” column, but is a more detailed indicator of the types of products that lead to the most profit. With this data, you can see where price points are failing to meet demand.
From these results, a business owner can see they get the most profit from copiers and are losing profit from bookcases and tables.
This data is also great for customer segmentation. Business owners can see from which customers they receive most of their profit.
You can even see your profit on a state-by-state basis.
Takeaways and Other Insights You Can Get from this Data
This is an amazing way for businesses to explore profit drivers in order to cut out inefficiencies and experiment with price points. This dataset holds much more value than just profit.
With the other attributes, retail stores can also conclude where your market is concentrated, for example.
This is incredibly important to knowing where to allocate marketing to and other company resources.
Other things you can predict and analyze with this dataset.
- Most common shipping method
- Average order priority by product sub-category
- Average time between order data and shipping date
- Quantity related to location and product
You can start getting the answer to these questions immediately by entering the Data Store.