数据集依据与“新冠肺炎”相关的230个主题关键词进行数据采集，抓取了微博数据，并人工标注.zip

# Covid-19-Analysis ### 1. Covid-19 Human Behaviour Analysis: A creative Reseach ##### (A) Description: - Herd Behavior is commonly used in Financial Markets to predict uncertainty. - Common Models : Econometrics Regression and Time Series - Research Value : Very Rarely applied in the Covid-19 Mobility case but Human behavior uncertainty affects the policy effectiveness #### (B) Model: - Baseline Model : Bayesian Ridge Regression --- ### 2. Covid-19 Sentiment Analysis: A Chinese NLP Study ##### (A) Description: - The dataset was based on 230 keywords related to the topic of COVID, and a total of 1 million blogs were collected from 1 January 2020 to 20 February 2020. THis includes sentiments : 1 (positive), 0 (neutral) and -1 (negative) (数据集依据与“新冠肺炎”相关的230个主题关键词进行数据采集，抓取了2020年1月1日—2020年2月20日期间共计100万条微博数据，并对其中10万条数据进行人工标注，标注分为三类，分别为：1（积极），0（中性）和-1（消极）) - 评价指标 Evaluation Matrix: Macro-F1 score #### (B) Model: - Preprocessing：tokenisation, remove duplicate, remove similar words - Machine Learning Model: Naive Bayes, Logistic regression -> result F1 score : 0.74 - Deep Learning Model：RNN (RNN+LSTM)， CNN