DataScienceProject

DataScienceProject (Code in Master Branche)

Association and prediction of daily corona virus infection with vaccination accumulations.

Theme: Vaccination, Covid, Health

Overview: This project aims to discover the relationship between coronavirus inflection with the accumulated vaccine doses for each day begin at December 14, 2020 to when I collected the data October, and try to predict the infection based on model. This project used two linear regression model, one multi-linear model build on public available dataset to predict how many vaccinations were needed in order to reach 0 infection. Tools used include seaborn, pandas, numpy, matplotlib, linear model from sklearn.

Data Sources:

https://github.com/nychealth/coronavirus-data/blob/master/trends/cases-by-day.csv, public available dataset from nychealth recorded the coronavirus infection in NYC.

https://github.com/nychealth/covid-vaccine-data/blob/main/doses/doses-by-day.csv, public available dataset from nychealth recorded the adminstrated vaccination per day in NYC.

WorK Done

1, Creating visualization for trends of virus infection and vaccination by using lineplot from seaborn, and malplotlib.

2, Data cleaning on two datasets so it can merge by using pandas.to_datetime(), dt.strftime(), pd.dataframe.merge().

3, Creating 2 linear models, and 1 multi-linear model for prediction on case infection by using Sklearn, pandas, study materials on linear regresion, correlation.

4, Adding attributes of prediction, error to existing data frame for residual graphs with the pd.series.apply(), pd.series substraction.

5, Creating visualizations for linear regressions and multi-linear regression by using sklearn, seaborn, malplotlib.

6, Creating residual graphs for corresponding model with seaborn, malplotlib.

7, Creating visualization for predict value pairs of cumulative dose2,1 to reach 0 infection based on multi-linear model with sklearn, seaborn, matplotlib, and algebra.

Citation

https://stackoverflow.com/questions/28371674/prevent-scientific-notation-in-matplotlib-pyplot https://stackoverflow.com/questions/60252480/how-to-plot-3d-multiple-linear-regression-with-2-features-using-matplotlib https://github.com/nychealth/coronavirus-data/blob/master/trends/cases-by-day.csv https://github.com/nychealth/covid-vaccine-data/blob/main/doses/doses-by-day.csv

Visualizations

The first two graphs show the trends of Positive covid case, and cumulative vaccine doses of dose1,dose2 from the time these data made available to October 2021. And the introduction of vaccine did drop the infection down dramatically. This suggest that vaccination and infection worth investigating.

The Linear Model with cumulative Dose 1 as independent variable to predict case count of corona virus infection.

This model predict that when Dose 1 cumulative reach around 6,062,831 (currently 6,152,675 on 11/08/2021), the corresponding corona infection would drop to 0. Only 1694 population increase on first dose cumulative will bring down 1 infection on average in this case. In addition, vaccination of dose1 and daily case count are highly correlated.

The error for this model is huge and having a pattern which denote the inability for this model to predict correctly.

The Linear Model with cumulative Dose 2 as independent variable to predict case count of corona virus infection.

This linear model predict that when Dose 2 cumulative reach around 5,912,397 (currently 5,476,764 on 11/08/2021), the corresponding corona infection would drop to 0. Only 1653 population increase on second dose cumulative will bring down 1 infection on average in this case. In additon, vaccination of dose2 and daily case count are highly correlated.

The error for this model is also huge and having a pattern which denote the inability for this model to predict correctly.

Since the residual graphs above suggested linear model may not make a good prediction, multi-linear model were, thus, build to see how well it can predict.

The multi-linear model with cumulative dose1/2 as independent variables, and virus infection case count as dependent variable

Like other two models, the residual graphs suggest there are other factor other than vaccine to affect the infection in NYC. One can see such pattern from all sudden rise of infection in August, September in the residual graphs.

One possible pair is (5470000, 6602608) which close to the current number of vaccination (5476764, 6152675) recorded in November 8, 2021. Base on this prediction, approximately 449,933 more people need to get first dose of vaccine in order to reach 0 infection in NYC.