From Zooming clients to neighborhood WhatsApp groups, digital platforms have become the only way for many of us to work, shop, get fit, or be educated. The physical analog world is being decimated. The digital world, however, is thriving. One of the responses we've seen to how people are approaching this period of isolation is in huge overnight changes to their shopping behaviors. From bulk-buying to online shopping, people are changing what they're buying, when, and how. According to a Statistics Canada report, Online shopping has doubled during the pandemic. This massive change can be good news for MSMEs since they can easily track and analyze their customers' behaviors with the help of Machine learning models.
What it does
In the Simplified ML app, we developed a web app with the help of the most known programming languages in the Data Science field (Python) that helps MSMEs train different ML models and chose the most suitable model according to their data's nature to maximize their revenue. The user can easily upload the dataset into the web app and submit a query by one click in the first step. For our case study, we used a reliable dataset published by walmart.ca. The web app will break the dataset into training and testing splits and deploy six different ML models in the next stage. (Regression classification model, Linear discriminant analysis, K-Nearest Neighbors, Classification and Regression Trees, Naive Bayes Classifier) While various datasets can respond differently to the ML models, we will measure the mean accuracy and standard deviation to find your dataset's best-fitted model. In this step, you can also see the highlighted features from the dataset in our app. At the next level, we visualize the result on our dashboard to forecast your total revenue in your next quarterly release. This web app can also help you maximize your profits.
How I built it
The web app was built in Python and Flask and it uses the SQLite as its database.
Challenges I ran into
Considering the customer's privacy and security challenged this project. This issue lead us to develop a local database on the user's end instead of using clouds or other available options.
What I learned
I learned how to code in Flask from 0 to what you are seeing for this hackathon.
What's next for Simplified ML App
- Adding some new ML models
- Improving the dashboard features.