Skip to content

praveendecode/IITM_ML_Project

Repository files navigation

Retail Sales Forecasting

image

Overview :

The challenge is to forecast department-wide sales for each store in the upcoming year, considering the influence of weather conditions, fuel prices, markdowns, and economic indicators. The objective is to decipher patterns, predict future sales, and derive actionable insights for strategic decision-making

Problem Statement:

  • Predict the department-wide sales for each store for the following year
  • Model the effects of markdowns on holiday weeks
  • Provide recommended actions based on the insights drawn, with prioritization placed on largest business impact

Main Features:

Data Collection:

  • Acquired data from Kaggle, comprising three different datasets.

Data Preprocessing:

  • Merged datasets using Python.

  • Utilized ML models to fill missing values.

Exploratory Data Analysis (EDA):

  • Conducted detailed analysis at store, department, and feature levels.

  • Explored patterns in markdowns, promotions, economic indicators, and weekly sales.

EDA and Tableau Dashboard:

  • Explored insights using Python.

  • Visualized key trends and relationships in a Tableau dashboard.

Model:

Model Selection :

  • Selected the best model among multiple model

Model Building :

  • built predictive models for sales forecasting.

Model Deployment:

  • Deployed the model using Azure, Docker, and Streamlit for user-friendly access.

Report:

  • Shared comprehensive PowerPoint reports with stakeholders.

  • Presented findings and actionable insights based on observed patterns and relationships

Tools Covered:

  • Python
  • Azure
  • Docker
  • Streamlit
  • Tableau
  • NumPy
  • Pandas
  • Seaborn
  • Matplotlib
  • Tableau
  • Scikit-learn
  • Github
  • Colab
  • Powerpoint

Results:

  • Successfully forecasted department-wide sales, achieving enhanced business impact.
  • Identified key drivers influencing sales through feature importance analysis.
  • Deployed a user-friendly predictive model accessible via Azure, Docker, and Streamlit

This project integrates advanced tools and methodologies to address the complexity of retail revenue forecasting, ultimately empowering strategic decision-making for substantial business impact

About

Designed an end-to-end ML model pipeline, forecasting department-wide sales by accounting for holiday markdown effects, spanning data collection to inferencing.

Topics

Resources

Stars

Watchers

Forks