ANALYZING KENYA'S MACROECONOMIC INDICATORS WITH PRINCIPAL COMPONENT ANALYSIS AND HIERARCHICAL REGRESSION

Abstract

This research set out to examine Kenyan macroeconomic factors using a hierarchical regression
model and Principal Component Analysis (PCA). A combination of descriptive and correlational
research methods was used in the study. From 1970 to 2019, data for the 18 macroeconomic
indicators were culled from the Kenya National Bureau of Statistics and the World Bank.For all the
data analysis, the R programme was used.In order to lower the data's dimensionality, we used
Principal Component Analysis, which reduced the original data set matrix to Eigenvectors and
Eigenvalues. To find out whether the extracted components were excellent at forecasting economic
development, we fitted them to a hierarchical regression model and used R2 as our measure of
success. The study's findings showed that the first component was strongly connected with fifteen
original variables and explained 73.605% of the total variance. As an added bonus, the two
variables exhibited a larger positive loading into the second main component, which explained
around 10.03% of the overall Variance. The third component, which had a strong correlation with
only one of the initial factors, explained about 6.22 percent of the total variation. A p-value of
0.0001<5% indicated that the models were significant, while the first, second, and third models had
F statistics of 2385.689, 1208.99, and 920.737, respectively. The third model was deemed the most
effective predictor due to its 17.296 mean square error. The reliability of Principal Component 1 in
describing economic growth was higher than that of other models since it had the largest Variance
explained and the lowest p-value. As a result, we may anticipate economic development in Kenya
based on the following macroeconomic variables: monetary economy, trade and openness to
government operations, consumption, and investment