THE GOOD GROCERY: New Product Sales Predictions

Overview

The Sales Predictions project is my first exploration into data science. I was given a data set of product charactaristics and sales in a number out outlets. The goal was to create a prediction method for future sales. The concept of future sales was not difined and given my interest in product development and industrial design i chose to use this as an opportunity to explore new market opportunities as well the concept of canabalism of sales. In the end my goal was to develop a ML model that would help the companies staff identify new products and predict their sales potential based on known charactaristics of any product.

Dashboard: INITIAL CONCEPT

Following the overview, my goal of this project was to create a new product model that would help the client assess new products and better understand their sales potential vs the existing products in their stores.

File Organization

|--README.md
|--main.ipynb
|--main_model.ipynb
|--templates
    |--index.html
    |--custom.css
    |--theme.yaml
|--plots
    |--a list of plots in html and png exported from index.html and main.ipynb
|--data
    |--sql
        |--The start of a SQL database for new and old product reporting.
    |--Data_Cleaning.ipynb
    |--outlet_item_plots.ipynb
    |--Outlet_Sales.ipynb
    |--outlet_totalsales.ipynb
    |--sales_app_data.py
    |--sales_predictions.csv
    |--sales_predictions_clean.csv

Tools Used

Pandas, MatPlotLib, Bokeh, and SKLearn.

The Data Cleaning + Exploration

You can see a scroll through of the index.html HERE

Plotting The Item Sales in Outlets

Creating the Prediction Model

All this work can be found in the main_model.ipynb. I will update this section with images and more text soon.

Resources

https://medium.com/y-data-stories/python-and-bokeh-part-iii-tutorial-116aa2e873eb

https://blog.hubspot.com/website/how-to-override-bootstrap-css

https://betterprogramming.pub/pickling-machine-learning-models-aeb474bc2d78

And many more