Labeled data pairs measurable inputs with known outcomes. In supervised learning, financial features such as prices, earnings, or macroeconomic indicators are paired with labels such as future direction, return, or cluster. section 8.1 section 8.1.2
Common labeling patterns include:
- Classification assigns categories such as buy, sell, hold, strong buy, or strong sell. section 8.1
- Regression assigns numeric outcomes, such as the percentage change in a stock’s price over one year. section 8.1
- Cluster labels can identify groups of companies with similar exploratory results, including a group associated with higher returns. section 8.1.2
A next-day price target
A stock table can use Open, High, Low, Close, and Volume as features, then create a Target column from the next day’s Close. A random forest trains on earlier rows, predicts held-out rows, and compares predictions with actual prices using RMSE and a plot.
