Goal

• PROJECT: Configure API pulls to various demographic (related) data sources, do some exploratory EDA, and trying to predict GDP using machine learning.
• PERSONAL: Get more confortable with ML and the project workflow.
  • This project is not meant to actually predict GDP lol

ML Problem

• Predict t+1 GDP per Capita (quantitative)
• The data:
  • large number of features
  • small number of rows

Steps

1. get data.ipynb: download the datapoints for each country from WorldBank's API and export data to a csv
2. preprocess data.ipynb: format data and filter out some features
3. model selection.ipynb: test various models' prediction error

Takeaways

• Part of ML is art, another part is just testing algos to see which works best and for what
• Two next projects:
  • datasets for testing: downloaded data and synthetic data generation with varying types of data and distributions
  • framework to plug in a model and test against various data types and distributions to evaluate strengths and limitations of models
    • goal: have an easy to plug into framework for baseline testing models on datasets, which should be beneficial for future ML projects from a knowledge and model selection perspective