AI-model on Detect Possibility of being covid-19 Positive and Negative

 

Step-by-step, making an AI-model following AI-project cycle;

Problem

Using 4Ws, lets answer this question:--

What is the problem?

--The problem is, when people has very less symptoms/common symptoms they go to the hospital for covid-19 test and in India in various states covid-19 charges is more than 2000/-, and if the test result come Negative, they waste their money and time.

Why you want to build this model ?

--This model will help those people who has very less symptoms and get worried about being covid-19 positive and can save their money and time. This AI-model will show them the probability of being positive and negative. So, based on probability they can decide whether to go for test or not.

Who will get benefited by this model ?

 --This model will help everyone during this pandemic all around the world and also reduces the stress on hospital from testing.

Where this model is applicable ? 

--This model is applicable in Medical fields or you can say health related sectors all around the world.

Data Acquisition

Covid-19 related data is available on On github and also on data.gov.in. I recommend you to download data from github as the data is converted into numerical format which is friendly language for algorithms.

Data Exploration 

Install all the required package / libraries :-

 import your downloaded data--

In data exploration we have to follow the below given instructions:-- 

• convert data into numerical format.
• use 0,1,2...... when you are dealing with categorical output.
• delete all the null values rows.
• or fill the null values with some other relevant values.
• remove all the erroneous data
  

CSV file of covid-19 data:--

 

This file do not contain any null values but in place of null vales it contains NONE so to deal with it follow the following step :--

• press CTRL + F .
• replace None with Blank space.

Modelling 

In this you have to use supervised learning algorithm named KNN (K-Nearest-Neighbors) for making prediction from given labelled data you can also called it as parameter-1.

Evaluation

This part is one of the most important part when you are predicting an output. In this you have to train and create your own Neural Network. This will give you probability of being positive and negative. While writing the code and training you have to ensure Over-fitting. train_test_split will do it.

Click here to get code