Home

POSHMARK -Assignment

Regression Problem

Identify/predict/recommend the price for given set of items

input :

x1-15

output (target) :

y_var

Data science Steps:

- Data Clean 
- Visualize the data
    - Ask Questions
- Feature Selection based on visualization and try brute force also
- Modeling:
    - Linear Regression
    - polynomial Regression
    - Gradient Boosting
    - Xgboost
    - Catboost
    - LightGBM
    - H20
    - Try Formula (if target was not fitted properly)

Deployment :

You can assume 5000 listings are being created in our platform every minute through our iphone 
or android app or website.

- Scalable solution 
    - should be Horizontally scalable
- Use load Balancer
    - async
    - docker
    - RPC call
- Use message Queue
    - KAFKA
    - Redis
    - celery


- Store the output of the model and data in some Big data systems (HDFS, S3, HIVE)

- should not have single point of failure
- Analytics as to be done
- Feedback based training
- Batch analytics

Explanation:

### Give explanation for your prediction if needed - lime - SHAP

Give confidence interval also for the price

Feedback:

- Take Feedback about the price recommendation
- Use this Feedback to improve/train the models

Attach Rest API

-Docker Deployment scripts