Forecasting Store Sales for Improved Decision-Making Using Machine Learning for Time Series Data
Accurate forecasting of sales is crucial for businesses across various industries as it provides several tangible benefits that contribute to improved decision-making and overall business performance. Some of the key ways in which accurate sales forecasting can positively impact a business include: Optimized inventory management; efficient allocation of resources such as human resources, production capacity, marketing budgets; Optimized production schedules; Effective marketing strategies; etc.
The AIM of this project is to build a model that more accurately predicts the unit sales for thousands of items sold at different Favorita stores.
Corporation Favorita is a large Ecuadorian-based grocery retailer
- Data Preprocessing
- Exploratory Data Analysis
- Modelling
- Model Evaluation
- Predictions
- Visualization
- PowerBI deployment
- Sphinx documentation generator
- Code Structure: Cookiecutter data science.
To set up the project locally and reproduce the results, follow the steps below: First navigate to the directory to be used for the project, open the terminal(you can type "cmd" in the path bar to open terminal from the folder):
- Create a new python virtual environment:
python -m venv *venv_name*
- Activate your venv_name :
.\venv_name\Scripts\activate
-
clone project repo with git clone command
-
Install the dependencies in the requirement.txt
(venv_name) pip install -r requirements.txt
- Now you can experiment with the codes. Refer to the references directory to understand folder struture.
To predict sales for a new horizon, follow the steps below:
Reach out at [email protected] for dataset. NB: dataset not uploaded to github due to size. both raw data and preprocessed data may be obtained by sending me an email.
List the main technologies, frameworks, and libraries used in your project.
- Programming Languages: Python, myst
- Documentation tool: Sphinx
- Data analysis and manipulation: numpy, pandas
- statistical modelling: statsmodel
- ML: Scikit-learn
- Data visualization: matplotlib, seaborn
Tietaar Louis, Brian Bassey, Umar Fawaz, Cornelius Cobbina.
This project is licensed under the MIT License.
Read an article on this project here: https://medium.com/@yebsolomon70/forecasting-store-sales-using-machine-learning-for-time-series-0a8d164b0626
-
Peng, R. D., & Matsui, E. (2015). The Art of Data Science. Skybrude Consulting, LLC
-
vishwas, B. V., Patel A. (2020). Hands-on Time Series Analysis with Python. Apress