Skip to content

A Machine Learning Product for Understanding the Macroeconomic Environment

Notifications You must be signed in to change notification settings

elainechowqz/Predictor-Macroeconomics-and-Business-Cycles

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

54 Commits
 
 
 
 

Repository files navigation

Macroeconomics and Business Cycles Predictor

This ongoing Python project analyzes over half a century of macroeconomic data (from the 1940s to the present, including the COVID-19 pandemic) to better understand changes in macroeconomic conditions and business cycles. Our analysis currently focuses on two key indicators: unemployment and inflation rates, which have historically shown interesting relationships. We've generated various plots to visualize the data and uncover patterns.

Using the K-Means Clustering algorithm, we’ve categorized the normalized unemployment and inflation data into three key clusters:

  • Low inflation, low unemployment

  • High inflation, low unemployment

  • Low inflation, high unemployment

Other clustering configurations also reveal statistically interesting groupings. These clusters are visualized to provide a clearer understanding of macroeconomic shifts.

3 clusters

Key Discovery

We discovered that changes in macroeconomic conditions, indicated by shifts in unemployment and inflation, follow a Markov process. Hypothesis testing revealed that the time between transitions between clusters (e.g., from "low inflation, low unemployment" to "low inflation, high unemployment" and then to "high inflation, low unemployment" during the COVID-19 pandemic) adheres to a Geometric distribution (or Exponential distribution in continuous time). This time interval is linked to business cycle lengths and the duration of various macroeconomic environments. Understanding these transitions enhances our ability to predict shifts in macroeconomic climates and business cycles.

Below are the outputs and potential business applications of our predictor.

Outputs

  1. Macroeconomic Environment Prediction

The model predicts the timing and likelihood of macroeconomic shifts and provides insights into the duration of business cycles. This has potential applications to business planning, like product pricing and budget planning.

  1. Other Potential Applications

Given the strong connections between macroeconomic indicators and financial measures like market returns and credit risk, this project opens up opportunities to extend the analysis to other business areas like financial risk management:

Predicting stock market returns under different macroeconomic conditions

Building regression models between macroeconomic indicators and credit default probabilities, with potential applications in predicting credit risks

About

A Machine Learning Product for Understanding the Macroeconomic Environment

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages