Description: I would like to create a project that analyzes Star Market product data in order to help me make optimal grocery shopping decisions.
Goal(s): Successfully predict the Star Market weekly ad sale prices of various food products.
Data Collection: The archive of Star Market weekly ads are available online through third-party websites. They go back a year, giving 52 weeks of data plus however many present/future weekly ads I want to collect. The archived ads are in image format and the data isn’t readily available or easy to extract. I’ll still try to automate it, though falling back on manual extraction shouldn’t be too bad. The data itself for individual items would include things like: name, price, discount rate, calorie count, food category, other nutrition info, etc. For the weeks, it might be worth considering things like season or holidays.
Data Modeling: Not too sure yet. But from vibes, I’m assuming that the weekly ad discounts should follow some regular weekly pattern.
Data Visualization: Not too sure yet. I’ll probably want a simple scatter plot where each point represents the presence of a discount for a certain food item across all of the weeks from the collected data. I also want to visualize the probability distributions of a given food item being on sale in the next week.
Test Plan: Testing on 4-5 instances of the weekly ads toward the end of October through November sounds good to me.