You only need to change one part of the code from the following link and get results. (Code and data are almost the same except for the training period.)
https://colab.research.google.com/drive/1by_LYqIXw8Z_l7rXfAlbfFIKNzo3Dpcf (Links to an external site.)Links to an external site.
You need to save a copy in Drive (File – save a copy in Drive) and work in your own colab page. You need to save the file and upload it in Colab.
Train samples are years 1961 to 2005, and test samples are years 2006 to 2019.
In each set, please choose 5 predictors among the following predictors:
FED_FUNDS_RATE_L1, INFLATION, OUTPUT_GAP, POPULATION_GROWTH_1Y, UNEMPLOYMENT_RATE, UNEMPLOYMENT_GROWTH_1Y, LONG_TERM_INTEREST_RATE, Price to Earnings Ratio, STOCK_TOTAL_RETURN_PRICE, STOCK_RETURN_1Y, CO2_EMISSIONS, TEMPERATURE, TEMPERATURE_DEVIATION, GDP, GDP_GROWTH_1YEAR, Shiller’s Price to Earnings Ratio, Dividend Yield, Nondurable Good Consumption Growth, Durable Good Consumption Growth, Michigan Sentiment Index, Producer Price Index Growth
where FED_FUNDS_RATE_L1 is 1 month lagged fed funds rate, ‘INFLATION’ is consumer price index (CPI), ‘OUTPUT_GAP’ is the difference between real gross domestic product (GDP) and real GDP, ‘POPULATION_GROWTH_1Y’ is 1 year population growth, ‘UNEMPLOYMENT_RATE’ is unemployment rate, ‘UNEMPLOYMENT_GROWTH_1Y’ is 1 year change in unemployment rate, ‘LONG_TERM_INTEREST_RATE’ is long term interest rate, ‘PE’ is price earnings ratio of aggregate stock market, ‘STOCK_TOTAL_RETURN_PRICE’ is total return price of aggregate stock market, ‘STOCK_RETURN_1Y’ is 1 year return of aggregate stock market, ‘TEMPERATURE’ is average temperature, ‘GDP’ is real GDP, ‘GDP_GROWTH_1YEAR’ is 1 year change in real GDP, CO2_EMISSIONS is average CO2 emission, and the rest are implied by their names.
For example, you can choose one set of predictors: X_list = [‘UNEMPLOYMENT_RATE’, ‘STOCK_RETURN_1Y’, ‘LONG_TERM_INTEREST_RATE’, ‘PE’, ‘TEMPERATURE’].
You only need to modify one part that defines X_list in the code and run the code. (It won’t take long time to finish the assignment.)
Describe the results of machine learning and compare the differences between the three sets of predictors to which combination is better in what sense. Please write the 4 pages (12 points, 1.5 line space, normal margin, excluding page cover) of reports including figures. Anything that you think is important in interpreting the results can be written. (For example, you write why some variable sets work better than other sets.) You don’t have to be too rigorous.
