hopefrost / hospital_stay_classifcation

Capstone Data Science Project

Springboard capstone project

hospital_stay_classifcation

Management in hospitals, holds a lot of responsibility and they need to have answers questions such as: How long will a patient be in the hospital? Will the hospital have enough resources to meet that patient’s needs? Every patient and their family will be intensely concerned with getting home and healthy. It is, however, imperative that Hospitals can plan for how long a patient will be admitted. HealthMan, a non-profit focused on the management and functioning of Hospitals, has posted a dataset on Analytics Vidhya. Hospitals have to constantly shuffle people and resources to meet every patient’s needs. With the recent pandemic, many people died simply because hospitals didn’t have enough resources. If the hospital knows how long a patient will need care they can plan their use of resources to save stress on patients and staff; while allowing them to optimize the care and treatments given.

This project is a multi-class classification of eleven 10-day ranges for length of stay in the hospital. The final models predictions rank in the top 1.5% of predictions submitted while the hackathon was open.

Project Files:

Data Wrangling and EDA:

1_cleaning_and_EDA.ipynb

Feature Engineering and Modeling:

2_features_and_modeling.ipynb

Data Folder:

Data Sets

test_data.csv train_data.csv

Data Discription

train_data_dictionary.csv

subission predictions

rf_submission.csv

Documentation folder:

Full Report

hospital_stay_prediction_report.pdf

presentation slides

hospital_stay_presentation.pdf