# Flipkart Smart Bag Brief Idea

## Fixed Effect
- Demographics
- Age group
- Staple Diet
- Brand Frequency
- **Recipe** (Identify Items in Foods in proportion to estimate monthly requirements)
- Spending Power
- Occasions

Festive Foods products will be tracked during festivals at a specific demographics. Offers will be provided.

## Random Effect
- Search
- Click
- Comment
- Rating
- Order History
- Health Data (Get health data from fitness app will help us recomend health based products. e.g. using weight data for high weight people we can recommend low protein foods. Using heart data we can recommend cholestrol free foods.)

## Create a smart basket for every user based on the following inputs:

### Time of visit


### Relevant products purchased by similar users


### Past Purchases


### Building for New users with no purchase history


### Identify repeat purchase products for all users along with the frequency with which they are repeating


### Identify groups of users who showcase similar buying needs


## Architecture


## Explanation

We will use k-means clustering to group customers into segments based on the products they have bought historically. We will implement K-Means on the share of units bought from the sum of each customer’s previous orders. We will perform PCA to reduce the number of features for the K-Means algorithm. This would allow us to better visualize our clusters as well as help K-Means run more efficiently.

We will use associative rule mining to get relationships products have with each other in terms of how likely they are to be bought together in the same order.
Three of the common rules are support, confidence and lift.
Products with highest similarities will have highest lift values.

## References

- https://towardsdatascience.com/building-a-food-recommendation-system-90788f78691a
- https://en.wikipedia.org/wiki/Fixed_effects_model
- https://en.wikipedia.org/wiki/Random_effects_model
- https://engineering.linkedin.com/blog/2020/gdmix--a-deep-ranking-personalization-framework
- https://blog.smile.io/how-to-calculate-purchase-frequency/