Developing a Literature Review for Machine Learning in Finance and Economics
Please check link to:
https://sites.google.com/view/vinegar...
Useful reading for a developing a Literature Review on Machine Learning with Finance Applications:
Big Data: New Tricks for Econometrics Hal R. Varian (2014)
https://www.aeaweb.org/articles?id=10...
Machine Learning: An Applied Econometric Approach, Sendhil Mullainathan, Jann Spiess (2017)
https://www.aeaweb.org/articles?id=10...
Machine learning methods in finance: Recent applications and prospects, Daniel Hoang, Kevin Wiegratz (2022)
https://onlinelibrary.wiley.com/doi/f...
A useful text book for Machine Learning (Finance)
https://www-2.rotman.utoronto.ca/~hul...
1. Objectives and Focus
Econometrics primarily focuses on understanding the relationships between variables, often with an emphasis on causal inference. It seeks to explain how one variable affects another and is concerned with the theoretical underpinnings of these relationships.
Machine Learning, on the other hand, prioritizes prediction accuracy and pattern recognition. The goal is often to develop models that can make accurate predictions about new, unseen data, rather than to understand the underlying causal relationships between variables.
2. Approach to Data and Modeling
Econometrics tends to emphasize model interpretability and simplicity, often starting with a hypothesis based on economic theory that guides the selection of variables and model structure. It uses statistical tests to validate models and hypotheses.
Machine Learning employs a more data-driven approach, where the model structure can be determined by the data itself. It often uses complex models that can capture intricate patterns in the data, sometimes at the expense of interpretability.
3. Techniques and Algorithms
Econometrics uses statistical methods such as linear regression, generalized linear models, and time-series analysis. These methods are designed to estimate the parameters of interest with an emphasis on unbiasedness, efficiency, and statistical inference.
Machine Learning encompasses a wide range of algorithms, including but not limited to decision trees, neural networks, support vector machines, and ensemble methods like random forests and gradient boosting. These methods are often more flexible and capable of modeling complex nonlinear relationships.
4. Evaluation Metrics
Econometrics models are often evaluated based on statistical measures such as r-square , t-statistics, F-statistics, and p-values that assess the significance of variables and the overall model fit.
Machine Learning models are evaluated using metrics that measure prediction accuracy, such as mean squared error (MSE), mean absolute error (MAE), accuracy, precision, recall, and the area under the ROC curve (AUC). The choice of metric depends on the specific prediction task and the nature of the data.
5. Assumptions
Econometrics models typically rely on strong assumptions about the data-generating process, including linearity, normality, homoscedasticity, and the absence of multicollinearity among explanatory variables.
Machine Learning methods often make fewer explicit assumptions about the data. They can handle nonlinear relationships and interactions between variables more flexibly, and they can adapt to various data distributions.
6. Applications
Econometrics is widely used in economics, finance, policy analysis, and other fields where understanding causal relationships is crucial.
Machine Learning has a broader range of applications, including but not limited to image and speech recognition, natural language processing, recommender systems, and autonomous vehicles, where predictive accuracy is key.