This project focuses on creating a procurement plan for the year 2023 based on a comprehensive analysis of medication purchase data from hospitals in Spain. It utilizes a detailed dataset containing information on product types, quantity demanded, and other relevant attributes for each hospital purchase.
healthcare_challenge
├── datasets
├── models
├── data_exploring
├── predictions
├── utils
- datasets: Contains the datasets used in the project.
- models: Stores the created models, including XGBoost ensembles, CatBoost, and predictive time series models like Prophet and ARIMA.
- data_exploring: Holds .py files used to explore and analyze the dataset.
- predictions: Includes the predictions generated by the various models.
- utils: Folder containing functions and utilities used throughout the code.
The project employed a variety of models for predictions and analysis, including:
- XGBoost: Ensembles created for different data cases.
- CatBoost: Employed for analysis and predictions.
- Prophet and ARIMA: Specific time series models used for predictive analysis.
The project has been organized into different directories for easy navigation and utilization of each part. It's recommended to follow the specific instructions in each folder to better understand the purpose and use of the files contained within.
For data exploration and a better understanding of its structure, the files in the data_exploring
folder can be used.
The employed models are found in the models
folder, with their respective predictions in the predictions
folder.
We're open to collaborations and suggestions! Feel free to contribute, whether it's new ideas, improvements to existing models, or the addition of new analyses.
We appreciate everyone involved in the development of this project and those whose work has been essential for its execution.
We are four Data Engineering students from UAB who are passionate about Deep Learning.
For any FAQs:
Note: Ensure you have the necessary dependencies to run the different scripts and models present in this repository. Refer to the specific documentation of each model or analysis for more details.