Explores and models the cost of groceries in different regions, investigating variations based on country, store brand, and potential correlations with rental prices.
In this project, I answered the following key questions:
- What is the average price of each product, and how do we anchor it considering different currencies and price distributions?
- How does the geographical location of grocery stores affect product prices, and how does the store brand classification (budget, mid-range, or luxury) influence these variations?
- Is there a correlation between price variation by geographical location and rental prices in the corresponding neighborhoods?
- Can we enhance our analysis by incorporating a spatially correlated distribution over location multipliers to reveal underlying patterns in price fluctuations?
To conduct the analysis, I collected price data for ten essential grocery items from various supermarkets across different countries. Each product has three recorded prices, accounting for brand variations. Additionally, I recorded the quantity of each product to standardize the prices.
To establish correlations, we sourced rental price data for the neighborhoods where these supermarkets are located.
I also explored the incorporation of a spatially correlated distribution over location multipliers. This modification allowed me to investigate hidden patterns in price variations, considering the spatial proximity of grocery stores.
I used PyStan to build the model. The model structure is based on a combination of a base price for each product and multipliers associated with store brands and geographical locations. This allows us to predict the final price of a product in a particular store effectively.