1 of 16

COVID-19

By: Sharmin Akther, Julia Drobotya, & Rita Wang

2 of 16

Reason for Selected Topic

  • COVID-19
    • We wanted to pick a topic in the Healthcare field that would seem beneficial to analyze for the community at this time
    • We selected the topic of COVID-19, because we are currently still in a situation where there are relatively high numbers of people with COVID-19

3 of 16

Questions the team hopes to answer with the data

  • Does the number of COVID-19 patients hospitalized correlate with the average percentage of COVID-19 patients who passed away during hospitalization?
  • Does the number of people who are tested positive for COVID-19 correlate with the number of deaths in the hospital?
  • Did the time of vaccination affect with the number of deaths in the hospital due to COVID-19?

4 of 16

Description of Data Source

  • The data was collected from NYC Open Data
  • The data source consists of the following:
    • Extraction date
    • Specimen date of COVID testing
    • Number of tested cases
    • Number of confirmed cases
    • Number of hospitalized
    • Number of deaths

5 of 16

Process of the Data Exploration

  • Cleaning data
    • Creating a Specimen Batch ID
    • Merging two tables using Left Join
    • Convert Speciment_Date data type to datetime

​

  • Uploaded to SQL database (Sqlalchemy)
    • Specimen Batch ID as primary key
    • Three tables

​

  • Machine learning Model was selected
  • Entity Relationship Diagram (ERD) created

6 of 16

Technologies, Languages, Tools, and Algorithms Used

Data Processing

  • Python
  • Jupyter notebook
  • Matplotlib

Database Storage

  • Postgres
  • Pgadmin4

​

​

​

Machine Learning

  • Scikit-learn

Visualization

  • Tableau
  • Plotly

Version Control

  • GitHub

​

7 of 16

Machine Learning Model

  • Supervised or Unsupervised?
    • Supervised → the data is labelled

​

  • Type of models used:
    • Linear Regression
    • Random Forest

8 of 16

Linear Regression

  • Positive correlation
  • Accuracy score → 0.97

Number of Hospitalized Patients

Number of Deaths

9 of 16

Random Forest Classifier

  • Accuracy score → 0.84�
  • Important features
    • Number_confirmed → 0.2968
    • Number_hospitalize → 0.2319
    • Number_tested → 0.25165

10 of 16

Covid-19 Peak Months

​

​

11 of 16

Confirmed cases vs. Hospitalized

Confirmed vs. Number of Deaths

​

12 of 16

Confirmed cases (Vaccination Program)

13 of 16

Death Rate Significantly Drops After Vaccination Program Started

​

← Death trend 01/2020 - 09/2021

Death trend in last 15 days of September 2021 →

14 of 16

Answers to the Questions

  • Does the number of COVID-19 patients hospitalized correlate with the average percentage of COVID-19 patients who passed away during hospitalization?
  • Does the number of people who are tested positive for COVID-19 correlate with the number of deaths in the hospital?
  • Did the time of vaccination affect with the number of deaths in the hospital due to COVID-19?

15 of 16

Recommendations

  • Our data had limited columns, for further studies we need to get more information like Borough Name, Hospital Name, Age Groups, Gender
  • Get vaccination records so we can compare them with vaccination rates and number of death

16 of 16

Thank you !

​