Disclaimer: This repository is created for a user study.
A company comes to you in order to improve their product sales prediction. Currently they use a simple ARIMA model to do their sales predition and they reach a MAE of 49.25 sales per day on the given dataset. Find out how you can improve the prediction of the company.
Predicting the number of sales is highly important for sellers in order to make informed decisions, optimize resources and ensure the sustainability and growth of their business. Accurate sales predictions help in managing inventory levels, ensuring the right products are available at the right time, avoiding overstock or stockouts and reducing carrying costs. Sellers can also plan marketing and advertising strategies effectively, allocating budgets to the most promising channels and campaigns. Accurate predictions are hard due to fluctuating economic conditions, consumer behavior, competitive landscape and even unpredictable events can have a profound impact on sales figures. This dataset contains number of sales, expenses and other attributes to an online product.
- n_sales: Number of sales in a day
- n_searches: Number of searches on the website for the product
- discount: Product discount (%)
- n_returns: Number of returns
- ad_expenses: Ad expanses (€)
- review_score: Customer Score Review (between 1-5)
- cost_of_goods_sold: Expanses for goods and production for the product sold in a day
Here are a template notebooks for the analysis. Modify the notebooks as you need!
| Notebook | Colab Link |
|---|---|
| Data Analysis | Colab |