Inflation forecasting with Random Forest
(2022) NEKH03 20212Department of Economics
- Abstract
- The accuracy of inflation forecasts is, and has been, important for economic agents such as
governments, central banks, companies, and the general public. Historically it has mainly
been conducted with traditional statistical models that limits the usage of bigger datasets.
This thesis will examine the performance of the machine learning model called Random Forest
by forecasting Swedish inflation between January 2016 and January 2020. The forecasting
horizons will be 1, 3, 6 and 12 months and the data used will consist of 250 variables, including lags and growth variables. For all forecasting horizons, except 1 month, Random Forest was able to provide more accurate predictions compared to the benchmark tests. The model also proved to... (More) - The accuracy of inflation forecasts is, and has been, important for economic agents such as
governments, central banks, companies, and the general public. Historically it has mainly
been conducted with traditional statistical models that limits the usage of bigger datasets.
This thesis will examine the performance of the machine learning model called Random Forest
by forecasting Swedish inflation between January 2016 and January 2020. The forecasting
horizons will be 1, 3, 6 and 12 months and the data used will consist of 250 variables, including lags and growth variables. For all forecasting horizons, except 1 month, Random Forest was able to provide more accurate predictions compared to the benchmark tests. The model also proved to effectively select predictive variables from a extensive data set and could therefore be useful in further quantitative research. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9073925
- author
- Meuller, Malte LU
- supervisor
- organization
- alternative title
- A Machine Learning approach to macroeconomic forecasting
- course
- NEKH03 20212
- year
- 2022
- type
- M2 - Bachelor Degree
- subject
- keywords
- Inflation, Forecasting, Random Forest, Machine Learning, Macroeconomics, Sweden
- language
- English
- id
- 9073925
- date added to LUP
- 2022-02-03 08:17:50
- date last changed
- 2022-02-03 08:17:50
@misc{9073925, abstract = {{The accuracy of inflation forecasts is, and has been, important for economic agents such as governments, central banks, companies, and the general public. Historically it has mainly been conducted with traditional statistical models that limits the usage of bigger datasets. This thesis will examine the performance of the machine learning model called Random Forest by forecasting Swedish inflation between January 2016 and January 2020. The forecasting horizons will be 1, 3, 6 and 12 months and the data used will consist of 250 variables, including lags and growth variables. For all forecasting horizons, except 1 month, Random Forest was able to provide more accurate predictions compared to the benchmark tests. The model also proved to effectively select predictive variables from a extensive data set and could therefore be useful in further quantitative research.}}, author = {{Meuller, Malte}}, language = {{eng}}, note = {{Student Paper}}, title = {{Inflation forecasting with Random Forest}}, year = {{2022}}, }