Description
Get up to speed on the application of machine learning approaches in macroeconomic research. This book brings together economics and data science. Author Tshepo Chris Nokeri begins by introducing you to covariance analysis, correlation analysis, cross-validation, hyperparameter optimization, regression analysis, and residual analysis. In addition, he presents an approach to contend with multi-collinearity. He then debunks a time series model recognized as the additive model. He reveals a technique for binarizing an economic feature to perform classification analysis using logistic regression. He brings in the Hidden Markov Model, used to discover hidden patterns and growth in the world economy. The author demonstrates unsupervised machine learning techniques such as principal component analysis and cluster analysis. Key deep learning concepts and ways of structuring artificial neural networks are explored along with training them and assessing their performance. The Monte Carlo simulation technique is applied to stimulate the purchasing power of money in an economy. Lastly, the Structural Equation Model (SEM) is considered to integrate correlation analysis, factor analysis, multivariate analysis, causal analysis, and path analysis. After reading this book, you should be able to recognize the connection between econometrics and data science. You will know how to apply a machine learning approach to modeling complex economic problems and others beyond this book. You will know how to circumvent and enhance model performance, together with the practical implications of a machine learning approach in econometrics, and you will be able to deal with pressing economic problems. What You Will Learn Examine complex, multivariate, linear-causal structures through the path and structural analysis technique, including non-linearity and hidden states Be familiar with practical applications of machine learning and deep learning in econometrics Understand theoretical framework and hypothesis development, and techniques for selecting appropriate models Develop, test, validate, and improve key supervised (i.e., regression and classification) and unsupervised (i.e., dimension reduction and cluster analysis) machine learning models, alongside neural networks, Markov, and SEM models Represent and interpret data and models Who This Book Is For Beginning and intermediate data scientists, economists, machine learning engineers, statisticians, and business executives Tshepo Chris Nokeri harnesses big data, advanced analytics, and artificial intelligence to foster innovation and optimize business performance. In his functional work, he has delivered complex solutions to companies in the mining, petroleum, and manufacturing industries. He initially completed a bachelor’s degree in information management. He then graduated with an honors degree in business science at the University of the Witwatersrand on a TATA Prestigious Scholarship and a Wits Postgraduate Merit Award. He was unanimously awarded the Oxford University Press Prize. He has authored two Apress books: Data Science Revealed: With Feature Engineering, Data Visualization, Pipeline Development, and Hyperparameter Tuning, and Implementing Machine Learning for Finance: A Systematic Approach to Predictive Risk and Performance Analysis for Investment Portfolios. Chapter 1 Introduction to Econometrics This is the preliminary chapter of the book. It covers the application of data science practices in econometrics. Sub-topics Econometrics Economic design Comprehending statistics Learning modeling Deep learning modeling Structural equation modeling Macroeconomic data source Context of the book Practical implications Chapter 2 Univariate Consumption Study Applying Regression This chapter introduces a simple linear regression model known as the ordinary least-square model. It applies the model to determine whether changes in lending interest rate (%) influence changes in final consumption expenditure (current US$) in the USA. It contains ways of conducting covariance analysis, correlation analysis, model development, cross-validation, hyperparameter optimization, and model evaluation. Sub-topics Context of this chapter Theoretical framework a) lending interest rate (%) b) Final consumption expenditure (current US$) The normality assumption a) normality detection Descriptive statistics Covariance analysis Correlation analysis The Pearson correlation method Ordinary least squares model development using statsmodels Ordinary least squares model development using SciKit-Learn a) Cross-validation b) Predictions c) Intercept and coefficients estimation d) Residuals e) Other ordinary least-square regression performance metrics f) Learning curve Conclusion Chapter 3 Multivariate Consumption Study Applying Regression The preceding chapter carefully covered simple linear regression-a model for predicting continuous response variables using a predictor variable. There are cases where there is over one predictor variable. This chapter presents ways of properly fitting multiple variables into a regression equation. It applies the ordinary least-square model to examine whether changes in social contributions (current LCU), lending interest rate (%), and GDP growth (annual %) influence changes in final consumption expenditure (current US$). First, it applies the Pearson correlation method to study the correlation among the variables, and then it implements the Eigen matrix to determine the severity among variables. Sub-topics Context of This Chapter a. Social contributions (current LCU) b. Lending interest rate (%) c. GDP growth (Annual %) d. Final consumption expenditure (Current US$) Theoretical framework Descriptive statistics Covariance analysis Correlation analysis Correlation severity detection Dimension reduction Ordinary least squares model development using statsmodels a. Residual analysis Residual autocorrelation Ordinary least squares model development using sciKit-learn cross-validation a. Hyperparameter optimization b. Residual analysis Residual autocorrelation a. Learning curve Chapter 4 Forecasting Growth This chapter covers a time series analysis model recognized as the additive model to forecast future instances of future GDP growth (annual %) in the U.S. Before implementing the model, it first discusses time series analysis assumptions, thereafter it covers tests for stationarity, white noise, and autocorrelation and different models for time series analysis. Sub-topics Descriptive statistics Stationarity detection Random white noise detection Autocorrelation detection Different univariate time series models a. The autoregressive integrated moving average b. The seasonal autoregressive integrated moving average model c. The additive model Additive model development a. Additive model forecast Seasonal decomposition Chapter 5 Classifying Economic Data Applying Logistic Regression This chapter introduces a binary classification method recognized as logistic regression. To begin with, it covers descriptive analysis, covariance analysis, correlation analysis, correlation severity analysis, and dimension reduction. Following that, it exposes a viable way of binarizing a continuous variable. Next, it employs the sigmoid function to operate an urban population, GNI per Capita, Atlas Method (Current US$), GDP growth (annual %), then predict decreasing and increasing life expectancy at birth, total (years) in the USA. Last, it hands out ways of analyzing model performance using the confusion matrix, classification report, ROC curve, Precision-Recall curve, and learning curve. Sub-topics The multicollinearity problem Context of this chapter Theoretical framework a. Urban population b. GNI per capita, Atlas method (current US$) c. GDP growth (Annual %) d. Life expectancy at birth, total (years) Outlier detection Descriptive statistics Covariance analysis Correlation analysis Correlation severity detection Binarize a continuous variable Dimension reduction Logistic regression model performance evaluation a. Confusion matrix b. Classification report c. ROC curve d. Precision-recall curve e. Learning curve Conclusion Chapter 6 Finding Hidden Patterns in World Economy and Growth This chapter introduces decision-making using the Hidden Markov Model. It applies the Gaussian Mixture model to identify hidden patterns in the world economy and growth to forecast future patterns. Sub-topics Application of the hidden Markov Model Descriptive statistics Gaussian mixture model development Graphically representing hidden states Order hidden states Chapter 7 Clustering GNI Per Capita on a Continental Level This chapter covers an unsupervised machine learning model for clustering known as the K-Mean. At the outset, it covers covariance analysis, correlation analysis, and dimension reduction on GNI per capita data of African countries. Next, it refers to an elbow curve to determine the number of clusters to include in the model. Last, it examines predicted labels and applies the Silhouette method to analyze the model’s performance. Sub-topics Context of this chapter Descriptive statistics Dimension reduction Cluster number detection Cluster results analysis K-Means model development a. Predictions b. Cluster centres detection K-Means model evaluation a. The Silhouette method Chapter 8 Solving Economic Problems Applying Artificial Neural NetworksThis chapter provides a high-level overview of deep learning. First, it covers a basic artificial neural network recognized as the Restricted Boltzmann Machine. In addition, it discusses the shortcomings of shallow networks and how modern artificial neural networks overcome them. Next, it encloses the Multilayer Perceptron and techniques for developing more complex artificial neural networks. Sub-topics Context of this chapter Theoretical framework Restricted Boltzmann machine a. Restricted Boltzmann machine classifier development b. Model evaluation Confusion matrix Classification report ROC curve Precision-recall curve Learning curve Multilayer Perceptron a. Multilayer perceptron model development b. Model evaluation Confusion matrix Classification report ROC curve Precision-recall curve Learning curve Building neural networks with Keras a. Network architecture b. Network wrapping c. Network training d. Model development Confusion matrix Classification report ROC curve Precision-recall curve Training loss and cross-validation loss across epochs Training loss and cross-validation loss accuracy across epochs Chapter 9 Inflation Simulation This chapter examines the impact of different scenarios of central government debt in Great Britain using the Monte Carlo simulation method. In particular, it employs this method to determine the probability of a change in the country’s central government debt across multiple trials. Sub-topics Understanding simulation Descriptive statistics Monte Carlo Simulation Model Development a. Simulation results b. Simulation distribution Chapter 10 Economic Causal Analysis Applying Structural Equation Modelling This chapter introduces a model for determining causal relationships between variables known as structural equation modeling. To begin with, it covers how one can frame a structural relationship, introduce a mediating variable to a structural equation, and develop a structural equation model. To conclude, it explores the technique of presenting model information, inspection, and reporting indices, and visualizing a structural relationship with a mediator. Sub-topics Framing structural relationships Context of this chapter a. Final consumption expenditure (current US$) b. Inflation, consumer prices (annual %) c. Life expectancy at birth, total (years) d. GDP growth (Annual %) Covariance analysis Correlation analysis Correlation severity detection Structural equation model estimation Structural equation model development Structural equation model information Structural equation model inspection Report indices Visualize structural relationship
