Research Concepts
Reliability & Validity
Accuracy vs. Precision
Bias vs. Variance Tradeoff
- Understanding the Bias-Variance Tradeoff
- As you introduce more variables to a model, bias tends to decrease. This means that on average the model will make predictions closer to the true value.
- However, at the same time, variance tends to increase with the added variables. This means that each time you input the model with another random set of data, the outputted predictions will vary a lot. Why? More variables in the model means you might be in danger of overfitting your specific data. This means your model is so specifically tailored to your data at hand that it probably won’t generalize well if you applied the model to other randomly drawn datasets.
Curse of Dimensionality
Correlation vs. Causation