Predicting_Economic_Growth
所属分类：时间序列预测
开发工具：Jupyter Notebook
文件大小：0KB
下载次数：0
上传日期：2022-02-15 20:22:43
上 传 者：
sh-1993
说明： ...数据库，其中包含所有流行的美国报纸的新闻。一个基于金融新闻标题训练的金融情绪模型，用于预测...，
(This project tests the hypothesis that signals in the form of sentiment from newspaper text can be used to materially improve the forecasts of macroeconomic variables including infaltion, GDP and unemployment rate. Our news corpus is drawn from GDELT database which contains news from all popular US newspapers. A financial sentiment model)
文件列表:
Data/ (0, 2022-02-15)
Data/DL_Data/ (0, 2022-02-15)
Data/DL_Data/test_data.csv (181478, 2022-02-15)
Data/DL_Data/train_data.csv (573401, 2022-02-15)
Data/DL_Data/validation_data.csv (143775, 2022-02-15)
Data/FRED/ (0, 2022-02-15)
Data/FRED/CPIAUCNS.csv (21814, 2022-02-15)
Data/FRED/GDPC1.csv (6076, 2022-02-15)
Data/FRED/UNRATE.csv (13302, 2022-02-15)
Data/GDELT/ (0, 2022-02-15)
Data/GDELT/CSV.header.dailyupdates.txt (893, 2022-02-15)
Data/GDELT/List.xlsx (161540, 2022-02-15)
Data/GDELT/gdelt.csv (276709, 2022-02-15)
Data/Headline_Trainingdata.json (189808, 2022-02-15)
Data/Headline_Trialdata.json (1979, 2022-02-15)
Data/Headlines_Testdata.json (64184, 2022-02-15)
Data/Microblog_Trainingdata.json (313230, 2022-02-15)
Data/Microblogs_Testdata.json (107523, 2022-02-15)
Data/all-data.csv (672006, 2022-02-15)
Data/sentiment/ (0, 2022-02-15)
Data/sentiment/sentiment_score_lr_0_48.csv (2423, 2022-02-15)
Data/sentiment/sentiment_score_lr_1500_2100.csv (2559, 2022-02-15)
Data/sentiment/sentiment_score_lr_2102_3101.csv (2624, 2022-02-15)
Data/sentiment/sentiment_score_lr_254_354.csv (2432, 2022-02-15)
Data/sentiment/sentiment_score_lr_355_636.csv (2483, 2022-02-15)
Data/sentiment/sentiment_score_lr_48_253.csv (2461, 2022-02-15)
Data/sentiment/sentiment_score_lr_637_903.csv (2477, 2022-02-15)
Data/sentiment/sentiment_score_lr_904_1500.csv (2556, 2022-02-15)
Data/sentiment/sentiment_score_lstm_0_48.csv (2422, 2022-02-15)
Data/sentiment/sentiment_score_lstm_1500_2100.csv (2517, 2022-02-15)
Data/sentiment/sentiment_score_lstm_2102_3090.csv (2569, 2022-02-15)
Data/sentiment/sentiment_score_lstm_254_354.csv (2430, 2022-02-15)
Data/sentiment/sentiment_score_lstm_3091_3101.csv (2418, 2022-02-15)
Data/sentiment/sentiment_score_lstm_355_636.csv (2467, 2022-02-15)
Data/sentiment/sentiment_score_lstm_48_253.csv (2450, 2022-02-15)
Data/sentiment/sentiment_score_lstm_637_9030.csv (2459, 2022-02-15)
Data/sentiment/sentiment_score_lstm_904_1500.csv (2517, 2022-02-15)
Data/sentiment_score_lr.csv (3193, 2022-02-15)
Data/sentiment_score_lstm.csv (3008, 2022-02-15)
... ...
# Predicting Economic Growth from Business News
The main goal of this project to investigate if sentiment from newspaper can be useful in predicting macroeocnomics variables (unemployment rate, GDP, and CPI).
There are three steps in running the result:
1. Data Creation and Model Training: project_create_data.ipynb; Shallow_ML_Models/Shallow_ML_Models.ipynb; LSTM_Model/LSTM_model.py
It will first create the data, and then traing the shallow ML model and also the Bi-LSTM model
2. Sentiment Scoring: project_sentiment_scoring.ipynb OR project_sentiment_scoring.py
This will count the number of article that have a positive/neutral/negative sentiment predicted by each model
3. Macroeconomics Forecasting: project_macro_model.ipynb
This will fit the macroeconomics variables with different model: the benchmark, and with various sentiment scores.
近期下载者：
相关文件：
收藏者：