Exploring the Relationship Between Democracy and GDP per Capita: A Neural Network Approach
This interdisciplinary project aims to investigate the correlation between democracy and GDP per capita by employing a neural network approach. By harnessing the capabilities of machine learning and integrating knowledge from diverse fields such as politics, economics, philosophy of free will, and cognition, the project seeks to illuminate the potential connections between democratic values and a nation's economic prosperity. Additionally, in the current era of rising totalitarian regimes, this work becomes particularly relevant as it offers potential solutions to address this global challenge .
The project starts by gathering comprehensive data from a diverse range of countries worldwide, covering multiple decades of development. The data preprocessing phase involves loading the dataset into pandas dataframes and extracting the relevant features and target columns: the degree of democratic values as the input and GDP per capita as the predicted value. To ensure consistency and comparability, we employ the MinMaxScaler from Scikit-learn to normalize both the features and target values. This normalization step allows us to effectively train and evaluate our neural network model .
For our neural network architecture, we choose a fully connected neural network, also known as a feedforward neural network or a multi-layer perceptron (MLP). This architecture, consisting of an input layer, hidden layers, and an output layer. The rationale for this choice lies in the MLP's ability to handle complex non-linear relationships and make accurate predictions.
The model training phase involves optimizing the model using the normalized training data. We employ the mean squared error (MSE) loss function and the Adam optimizer to minimize the difference between predicted and actual GDP per capita values. We monitor the loss value for each epoch to assess the model's performance, taking into account the MSE criterion .
Our findings indicate that democratic values can predict a country's GDP per capita, but we acknowledge limitations and suggest exploring additional variables to enhance accuracy. We emphasize the role of free will in shaping economic progress, individual agency, and societal well-being, recognizing GDP per capita as an indicator of prosperity.
 S. Levitsky and D. Ziblatt, How Democracies Die. New York: Penguin Random House, 2018.
 P. Virtannen et al., "SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python," Nature Methods, vol. 17, no. 3, 261-272, 2020, doi: 10.1038/s41592-019-0686-2
 D. P. Kingma and M. Welling, "Auto-Encoding Variational Bayes," arXiv:1312.6114, 2013, doi: 10.48550/arXiv.1312.6114