This notebook presents an in-depth analysis of healthcare patient satisfaction data from 2016 to 2020. The analysis focuses on understanding historical trends, identifying influential factors, exploring regional variations, and employing predictive modeling techniques to forecast patient satisfaction and quality measures.
Key findings include:
Historical Trends: Time series analysis revealed trends in patient satisfaction over the years (2016-2020), showing how average patient survey star ratings have evolved. Predictive Modeling: Predictive models were developed to forecast patient satisfaction. Feature importance analysis highlighted the most influential variables in these forecasts, such as Hospital Overall Rating, Number of Completed Surveys, and Survey Response Rate Percent. Regional Variations: Geospatial analysis was used to visualize regional variations in patient satisfaction across different states, providing insights into how geographical factors may contribute to healthcare outcomes and patient experiences. Statistical Significance: Various statistical tests (Pearson correlation, Spearman rank correlation, T-Test, Chi-Square Test, Logistic Regression) were employed to assess the significance of relationships and correlations between different variables and their impact on patient satisfaction and emergency services. Machine Learning Algorithms: Several machine learning classification algorithms were explored using PyCaret to predict patient survey star ratings, with emphasis on handling the multiclass nature and potential class imbalance of the target variable. Based on the analysis, several recommendations are made for healthcare policies and interventions aimed at enhancing patient satisfaction and the overall quality of healthcare services. These recommendations are derived from the statistical and machine learning model results, highlighting areas for improvement based on the identified influential factors and their relationships with patient satisfaction and emergency services.
The predictive models, particularly those identified as having high accuracy in the PyCaret comparison, can be valuable tools for healthcare administrators and policymakers to proactively identify areas needing attention and to implement targeted interventions to improve patient care and satisfaction.