Siddharth Deshpande

Analyzed the COVID-19 dataset, a collection of scientific papers related to COVID-19, using different NLP techniques. The aim of the project was to get different insights using different NLP algorithms that might help understand research papers better.

Research papers available on GitHub:

• AutoDetect_COVID_FakeNews - Classification model for detecting Fake news regarding COVID
• BERT_semantic_search - Semantic search which finds similar sentences in COVID corpus in response to query question
• Biorelated_sentence_extraction_COVID - extract bio-related sentences from COVID corpus
• COVID_19_topic_modelling_Top2Vec - Topic modelling on COVID_19 corpus using Top2Vec
• COVID_explore_drugs - Explore drugs in the COVID corpus
• CoVID19_Ques_and_Ans - Covid papers Questions and Answering system based on doc2vec
• CoVID_19_NER_text_summarization_and_topic_modelling - BART summarization and LDA topic modelling and NER
• Covid_19_genome_analysis - COVID_19 genome analysis
• Covid_paper_rank_display - NER and covid papers recovery based on topic
• Medical_NER_Corona - NER on coronavirus dataset
• Mining_COVID_keywords - mining keywords using bigrams and trigrams

Won an Innovation Award for the project.

The goal of my project was to analyze the effect of weather on energy generation and demand and find a solution that can predict renewable energy generation and energy demand using weather parameters.

SOLUTION HIGHLIGHTS

• Solar, wind, and hydro energy generation prediction using climate and time parameters.
• Energy demand prediction was done using time and energy parameters (Model 1) and time, energy, and climate parameters (Model 2). Model 2 showed slightly higher accuracy than Model 1. It shows that climate parameters do not affect energy demand as significantly as energy parameters.
• Energy price prediction was done using time and energy parameters (Model 1) and time, energy, and climate parameters (Model 2). Model 2 showed higher accuracy than Model 1. It shows that climate parameters affect energy prices significantly.

For all the above cases, 10 million regression algorithms were tested. The ExtraTreeRegressor algorithm showed the best performance and was used to build the regression model.

URL: https://www.alibabacloud.com/blog/project-showcase-%7C-effect-of-weather-on-energy-generation-and-demand_598252

I built three conversational chatbots across different conversation channels, including Slack, WhatsApp, Dashboard, Discord, Telegram, and Facebook messenger. I developed chatbots using GPT-3 with additional constraints and prompt engineering.
• Conversation helper - This bot helps to simulate tough conversations so that the clients can practice the conversations beforehand. The client is scored on 2-3 conversation skills, and a report is generated at the end that shows his score and how to improve his conversation ability.
• Fashion assistant - This bot recommends fashion items based on client needs and stock inventory of the business. It uses a combination of GPT-3 and DALL-E.
• Google bot - This bot has a Google search engine capability and acts as an advisor/friend to whom you can ask any questions, and it will run a Google search in the back end to provide you with the most updated answers.
Bot previews can be shown during interviews.