Ecuador Sentiment Analysis 2020 COVID-19
In the year 2020, it was a year that impacted the entire world, with a pandemic that hit many of our countries, it seems that it was taken from a science fiction movie. In 2019, we made plans or projections for 2020, but with the coronavirus, most of our plans had to change, even for people like students who had to adapt to online education, workers working from home, and always keeping in mind a mouth cover, but despite everything that has happened in 2020, people moved forward, opened paths where there was nothing before, and continued on their way. The human being is impressive. After this motivational introduction. The present article is the impact that the coronavirus has had in Ecuador.
Cases of coronavirus in Ecuador
To date, the National Institute for Public Health Research (INSPI) has taken 1,022,897 samples for RT-PCR COVID-19, of which 289,472 are confirmed cases with PCR tests.
This indicator, updated daily, reports the accumulated number of samples taken to carry out the RT-PCR test in authorized laboratories in Ecuador. It should be noted that there may be more than one sample per person during the diagnostic process.
— 247,898 recovered patients.
— 32,150 cases with hospital discharge.
— 11,157 people deceased (confirmed COVID-19)
— 816 stable hospitalized.
— 507 hospitalized with reserved prognosis.
— 679,026 cases were discarded.
To analyze the sentiments, we had to collect data from Twitter, for that we used a Python library called Twint. Twint is an advanced Twitter scraping tool written in Python that allows you to scrape Tweets from Twitter profiles without using the Twitter API. . Twint uses Twitter’s search operators to allow you to scrape Tweets from specific users, scrape Tweets related to certain topics, hashtags, and trends, or classify sensitive information from Tweets such as emails and phone numbers. I find this very useful, and you can also be very creative with it. Twint also makes special queries to Twitter, allowing it to also track a Twitter user’s followers, what tweets a user has liked, and who they follow without any authentication, API, Selenium, or browser emulation.
With Twint, approximately 143,000 tweets were collected from January 1 to January 31. For the investigation, only 1000 tweets were used randomly, the computational power that I present now was not enough to be able to analyze the entire Dataset.
Relevant Ecuador News
Within the Dataset, there were days that had more tweets compared to the others, and those dates coincide with this news in the equator.
2020–02–29
The first case of coronavirus confirmed in the country
2020–01–20 to 2020–01–27
There were rumors for several days within social networks that there were public officials who supposedly died with COVID-19, until they published the falsehood of the case
Word Cloud
Before starting with the feelings of the people, a Word Cloud was also developed with the tweets of the users, in order to know which words were more relevant in the Dataset
In the most relevant words are pandemic, cases, day, positive, Guayaquil. Guayaquil, in the first dates that the coronavirus was introduced in Ecuador, was one of the cities most affected by the coronavirus, and there was a high rate of infection and a large part of the population was terrified of the case, most of whom were tested Because he had symptoms, his result was positive for COVID-19.
Topics
Topics were also generated to see titles were important within those dates, for that LDA is used. Latent Dirichlet Allocation (LDA) is a generative model that allows sets of observations to be explained by unobserved groups that explain why some parts of the data they are similar. For example, if the observations are words in documents, it assumes that each document is a mixture of a small number of categories (also called topics) and the occurrence of each word in a document is due to one of the categories to which the document belongs.
Sentiment Analysis
As previously stated, 1,000 tweets were used for each month to perform the analysis. Here are the results:
With normalized data
Example of NEUTRAL Tweets
Example of NEGATIVE Tweets
Example of POSITIVE Tweets
Conclusions
Ecuador was a country that was greatly affected by the arrival of COVID-19, there were many infected, through the data, people’s feelings are more negative, it is logical, it is a situation that makes a country sad, schools are closed, centers schools, full hospitals, deaths, infected, but little by little we will get out of COVID-19. The investigation only carried out approximately 12,000 data or tweets, a deeper analysis can be carried out with more data. I hope to carry out the tests in the future.
