COVID-19 - INDIA ANALYSIS

EXPLORATORY ANALYSIS

This RPubs documented is created to learn more about the COVID19 Confirmed Cases for India. This link will be usually shared in my personal blog i.e.; http://www.arunkoundinya.com. For other elated blogs request to visit my blog page and post your comment there.

Hope you will love the analysis and also here is the source code in the github account. https://github.com/arun0710/COVID19 Request to post your any improvement on the code for my learning to my E-Mail ID

TimeSeries Interactive State-wise Map with Confirmed Cases

Hope you will loved this interactive chart on Rpubs instead on RShiny :). My Wife loved this timline bar a lot as it is just intuitive enough to know which states have comparitvely more cases than others.

Interactive State-wise LogScale & Linear Scale of Confirmed Cases as on Date

Below is a Linear & Log Scale Graph of the Confirmed Cases Over Time; The Log Scale Graph has became more popular to highlight the concept of flatenning the curve. If we start playing around these dropdowns we can draw more insights which state has started early corrections with preperation like Kerala.

Understanding the Relationship between the Tests & Confirmed Cases Over Time

Below graph communicates to us very clearly; How well Tests and Confirmed cases are correlated to a larger extend in Indian Sub-Continent.

Clusters based on Positivity Rate and Confirmed Cases

Below graph communicates to us very clearly; In Descending Order of Future HotSpot States except Maharashtra. The right most cluster indicates that the testing is not being done or reported.
From 5th July Delhi & TamilNadu States have been removed for this below cluster graph

PREDICTIVE ANALYSIS

Source code will not be uploaded into the github account for this tab. Request to post your any improvement for my learning to my E-Mail ID

Basic TimeSeries Forecasting on Tests with Regression on Confirmed Cases

As we learnt in India Scenario that more the tests more the confirmed cases. This is a base model which uses time-series forecasting model on Tests and then the output of tests is regressed to get the total confirmed cases.

Active Caes, Total Confirmed Cases, Total Deathson 15th day starting from today. As a rule based of current Asympotomatic, Mild Cases upon Hospitalized Cases. Required Hospitalized cases can be calculated as 30% of total Active Cases considering the same recovery rate for even hospitalized cases.

	Results_15thDay_ FromToday
Base Model Forecasting Summary on 15th Day from Today
Date _15thDay	2020-07-20
ConfirmedCases_Total _15thDay	1073461
ActiveCases _15thDay	384818
Deaths_Total _15thDay	26887
Hospitalized People _15thDay	115445
† Hospitalized People are asummed as 30% of Actice Cases

SIRS Model with Undected Cases with revised transmission probability

Initial parameters for the SIRS Model for the suspected pouplation is kept for all urban & metro cities and then by past trend it is seen that the testing is 20 times lower than expected. As SIRS Model is being considered as over optimistic the initial set variables play a huge role in terms of arriving at the Infected Inviduals.

Also, Active Cases is arrived as the present recovery trend instead to use SIRS Model.

Note: There might be few changes as we learn during the pandemic course. The suggested model would certainly undergo changes on weekly basis. Here i have assumed that there will be no second wave.

Quicky Summary on 15th, 30th & 40th Day Prediction for easy readability.

	Results_15thDay_ FromToday \|	Results_30thDay_ FromToday \|	Results_40thDay_ FromToday \|	Results_60thDay_ FromToday \|
Adjusted SIRs Model Forecasting Summary on nth Day from Today
Date	2020-07-20	2020-08-04	2020-08-14	2020-09-03
ConfirmedCases_Total	1340926	2292388	2651331	2709475
ActiveCases	652283	1358045	1534120	1199143
Deaths_Total	26887	35726	42459	57847
Hospitalized People	195685	407414	460236	359743
† Hospitalized People are asummed as 30% of Actice Cases

Thanks a lot,
Arun Koundinya