Predicting the Number of Sunspots Per Month and Per Quarter Using ARIMA Models

Started Aug. 11, 2022

Abstract or project description

The number of sunspots in a given year changes as the sun goes through solar cycles, with peaks happening at regular intervals. When these peaks are plotted, a curve appears, similar to the oscillating sinusoidal wave. Because of its oscillatory nature, predictions of future sunspot values could be found since it is safe to assume that the number of sunspots would always follow a pattern. However, a simple, ordinary sine function, or any algebraic function for that matter, would not allow us to plot and predict future data points due to the complexity of the curve at hand. This led us to the hypothesis that in order to predict the future number of sunspots, models that involve autoregressive and moving average components (namely the ARIMA model) would be the most effective. In order to measure effectiveness, the mean-squared error (MSE) would be used, with a lower value (closer to 0) meaning better fit. The reason why we chose these sophisticated models was because these models took into account prior data points and their trends and seasonality to predict future data points. This essentially meant that this model would predict based on prior points, not on a fixed point or equation, like the sine curve. After plotting all of these models and finding the MSE for each, we drew the conclusion that the ARIMA model proved to produce the most accurate curve, with a MSE of only 460, as compared to the MSE that the best sine curve could produce: 21 million.

Analyzing the distribution of energy sources in the United States

Started Nov. 8, 2022

Abstract or project description

Energy consumption in the United States is an important element of the United States, but its use and value change from state to state and from sector to sector. Each state has its own unique needs, resulting in drastic variations in the consumption of certain forms of energy. It can be seen that states like Wyoming, for instance, appear to be outliers in the consumption of certain energy sources, while some other states might have comparable amounts for those energy sources. We used three groups of data to develop our analysis: grouping each form of energy and seeing how its consumption changes over time, grouping energy consumption by state and seeing how it changes over time, and grouping energy consumption by sector and seeing how it changes over time. We used a bar chart to explore the changing data, in order to see how consumption changed over the years in the United States.