bitgrit Releases a Twitter Sentiment Analysis Dashboard Amid COVID-19 Pandemic
This proprietary data analysis tool helps data scientists better understand and analyze social implications of the novel coronavirus.
As the situation continues to worsen worldwide around the COVID-19 pandemic, bitgrit is eager to lend our online AI services however we can to help understand and thereby lessen the disease’s destructive impact on society. This is why we developed a Twitter sentiment analysis dashboard to be used by data scientists to better understand trends of how people feel about the virus.
This dashboard is capable of extracting public Twitter data and has several possible implications across industries. For governments, it can help measure societal acceptance and other sentiments of policy decisions. In the business sector, it can help forecast demand and possible job shortages. On the individual level, it can provide highly localized reactions to developments surrounding COVID-19 and even forecast potential stockpiling behavior.
To glean useful insights from Twitter sentiment data, the dashboard follows this four-step process:
- Step 1 Tweet Extraction
First, the dashboard intelligently collects tweets related to COVID-19 using an API that picks up specific hashtags.
- Step 2 Data Preprocessing and Storage
Because Twitter data is collected from a public source, it requires preprocessing to conduct downstream analysis. The platform dynamically and automatically cleans the data in Python using an NLTK package that uses techniques such as word tokenization, removal of stop words, and lexicon normalization. This data is then loaded into MySQL for storage.
- Step 3 Model Training and Exploratory Data Analysis
Next, the Twitter sentiments are analyzed from the structured dataset using TextBlob, and exploratory data analysis is conducted using Pandas and Seaborn to derive meaningful insights. This analysis is recorded in terms of polarity scoring, sentiment word frequency, sentiment word clouds, emotional categorization, and other methods. These are then compared with historical associated responses to forecast future impact.
- Step 4 Real-time Data Visualization
Lastly, the insights and forecasts are visualized on a time series through a real-time dashboard that is connected to Ploty and is deployed using a front-end web app using Dash & Heroku PostgreSQL on Heroku server.
Along with this tool, we at bitgrit would like to extend an invitation to all data scientists and engineers to join us on our Telegram in a discussion on how we can use this tool along with our AI competition platform to join the larger global community in our common aim to stop the destruction caused by this deadly disease.
Our online competition platform can host petabytes worth of datasets from any industry, of any type, and from anywhere in the world. Our previous AI challenges have covered financial, philanthropic, and agricultural topics using datasets consisting of satellite images, historical news, and currency exchange rates. We hope that our platform can help uncover answers to some critical questions in the uncertainty of today’s climate.
This is an open call to all data scientists, AI engineers, and others in the AI or related fields to join the conversation about how we can best use this new Twitter sentiment analysis dashboard and our AI competition platform to contribute to society in these trying times. Click here to join in the discussion on our Telegram, we look forward to seeing you there.
