Sales Prediction Algorithm for an Apparel Retailer

Project Description

Everester is a mid-sized Apparel retailer. They manufacture and sell a variety of apparel for women, kids, and men. They have stores mostly in the USA. We have the sales history data for sales at stores for the past 3 years. We also know the store locations to understand where these stores are located and how this might influence the sales at these stores. We need to predict the sales per day for each of the next 100 days. We like to have a data science program that can also look at adjacent factors that influence sale including location and weather. Having an effective sales prediction will allow the store to figure out when to advertise more vs. when to advertise less. Plus it also will help them plan the inventory and staffing levels at a store.

We want you to create a general purpose data science module in python or R that can 

- Digest the daily sales data for the past 3 years and develop a model.

- Identify and use any publicly available sources of data for influencers on sale.

- Produce a prediction model that is cheap to run and provides higher accuracy of sale (with 3% of tolerance).

- Use the model to produce predicted sale at the stores for next 100 days. 

- The model should run with minimal deployment steps by an engineer at Transform. 

- Stretch Area:  Recommend what days are good to do promotions and what days are not good to do promotions to invite more customers.

As part of the deliverable, we require a 2-4 page writeup of your findings, besides the code module.

Data Set: The data set and problem are proprietary and confidential. Please do not share this with others.

The data will be shared once we get through the initial problem acceptance phase. We will start with data for 3 stores. There will be about 2500 rows of data for 3 stores, provided as a CSV file of size of 200KB. And we expect to have Transform run the model for additional stores to validate the model. 

And here is a sample data file for the format

# Daily Sales Data

# Store: TB1 = Tacoma, WA; PO1 = Portland, OR; DE1 = Denver, CO

Date,RevenueUSD

4/24/16, TB1, 40183

4/24/16, DE1, 18849

4/24/16, PO1, 33413

4/23/16,TB1, 73952

4/23/16,PO1, 43945

4/23/16,DE1, 63149