=======

Optimal Sales Forecast Model Assessment - eCommerce Startup

Sales forecasting is challenging for any business because of the limited historical data, unprecedented external factors like weather, natural events, government policies and mostly because of the volatile nature of market.

This project aims to apply various forecasting method to do Sales forecast for Brazilian E-Commerce startup Olist.

1. The submission packet has two folders for data: data and data_cleaned.

• The data folder consists of originally downloaded data from Kaggle.
• The data_cleaned folder consists of six .csv files that are cleaned data files. It consists of a master table which is the final dataset I have used for modelling.

2. Environment_specs.txt file has the environment specification that has been used to write jupyter notebooks.
3. There are two notebooks

• Data_cleaning_notebook_1 is the first notebook which loads actual data, cleans and merge all the table into master table.
• Time_series_notebook_2 is the second notebook where I have tried different modelling approaches to time series forecasting.

4. There are two pdfs for report and presentation.

In addition to this work, I have also prepared and published visualizations in Tableau which include dashboards for business and product strategy heads along with some stories bringing out crucial insights about the popularity of products by month, the relationship between freight and rating, customer and seller location, etc.