Analytical Methods:
The analysis employed a combination of statistical, time series, and machine learning techniques to derive meaningful insights from the dataset.
- Time Series Analysis:
- Seasonal-Trend Decomposition using LOESS (STL): Decomposition of selected numeric columns (excluding ‘Year’ and ‘Month’) into trend, seasonal, and residual components. This method illuminates underlying patterns and trends over time.
- ARIMA (Autoregressive Integrated Moving Average): Utilized for forecasting specified numeric columns based on historical data, predicting future values.
- Long Short-Term Memory (LSTM) Networks: Implementation of LSTM networks for specified numeric columns. LSTM, a deep learning technique, is well-suited for sequence prediction problems like time series forecasting.
- Correlation Analysis:
- Pairwise Scatter Plots: Construction of visual scatter plots for selected numeric columns to visually analyze associations between variables.
- Correlation Matrix: Quantification of linear correlations between variables through the creation of a correlation matrix for selected numeric columns.
- Identification of High Correlations: Selection of variables with high linear correlations (above a predetermined threshold) and analysis of their correlation values.
- Insights from Scatter Plots:
- Construction of scatter plots to investigate correlations between pairs of columns with strong correlation, offering visual insights into potential economic indicator dependencies.
These methods facilitated a comprehensive examination of the economic indicator’s dataset, revealing temporal trends, forecasting capabilities, and interactions between variables. The integration of classical time series analysis with modern machine learning approaches provided a holistic view of the economic landscape reflected in the dataset.