Abstract

• This project aims to investigate whether there exist a significant difference in electricity demand between developed and developing countries. Our dataset is sourced from the “Our World in Data” kaggle page, covering countries from the 1900 - 2022 and their energy economies. Our focus for this project will be on the last 30 years (1990 - 2020). We used regression analysis to model the relationship between electricity demand and country development status. We also used t-tests and ANOVA to compare the mean values of electricity demand between developed and developing nations. Our results showed that there was a significant difference in electricity demand between the two groups, with developed nations having higher values for electricity demand. The results of this project provide insights into the state of electricity demand in developed and developing countries. The findings suggest that improving the economy in developing countries may lead to an increase in electricity demand. The results also highlight the importance of investing in renewable energy sources to meet the increasing demand for electricity in both developing and developed countries. The findings have important implications for policymakers, energy companies, and investors, as they provide a better understanding of the factors that affect electricity demand.

Introduction

• Research Question :

Do developed countries have a significantly different electricity demand than developing countries?

• Hypothesis Testing :
  • \(H_0\): Mean electricity demand is the same
  • \(H_1\): Mean electricity demand is not the same
• Observation :
  • Difference in mean is observed in the data, but is this statistically significant at \(\alpha\) = 0.05 ?
• Data Methodology :
  • Web scraped HDI data from https://www.theglobaleconomy.com/rankings/human_development/
  • Merged energy economy and hdi dataset
  • Classification:
    • “Developed” : HDI >= 0.8 & gdp per capita >= $18,750
    • “Developing” : Otherwise

Introduction

• Data Collection :

https://www.kaggle.com/datasets/pralabhpoudel/world-energy-consumption

https://www.theglobaleconomy.com/rankings/human_development/

Summary Statistics

Summary Statistics

Summary Statistics

Analysis and Visualization

• \(H_0\): electricity demand is equal

• \(H_1\): electricity demand is not equal

• \(\alpha\) = 0.05

Statistical Analysis

t.test(electricity_demand ~ status, data = prime_df)

## 
##  Welch Two Sample t-test
## 
## data:  electricity_demand by status
## t = 8.247, df = 1666.6, p-value = 3.257e-16
## alternative hypothesis: true difference in means between group developed and group developing is not equal to 0
## 95 percent confidence interval:
##  121.9851 198.1150
## sample estimates:
##  mean in group developed mean in group developing 
##                 237.5535                  77.5035

ANOVA

aov(electricity_demand ~ status, data = prime_df) |>
    summary()

##               Df    Sum Sq  Mean Sq F value Pr(>F)    
## status         1  20574690 20574690   84.92 <2e-16 ***
## Residuals   3963 960110844   242269                   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 1148 observations deleted due to missingness

Regression Model

reg_model <- lm(avg_electricity_demand ~ year + status, data = ml_data)
summary(reg_model)

## 
## Call:
## lm(formula = avg_electricity_demand ~ year + status, data = ml_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -59.108 -12.762  -3.931  11.225  61.249 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      -6737.4198   732.4478  -9.198 5.34e-13 ***
## year                 3.4704     0.3651   9.505 1.66e-13 ***
## statusdeveloping  -154.7933     6.5315 -23.700  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 25.71 on 59 degrees of freedom
## Multiple R-squared:  0.917,  Adjusted R-squared:  0.9142 
## F-statistic:   326 on 2 and 59 DF,  p-value: < 2.2e-16

Regression Model

Classification Model

Data Partition

trainIndex <- createDataPartition(class_df$status, p = 0.7, list = FALSE)

train_data <- class_df[trainIndex, ]
test_data <- class_df[-trainIndex, ]

Model Training

ml_model <- naiveBayes(factor(status) ~ gdp_per_capita + hdi_2021 + continent, data = train_data)

predictions <- predict(ml_model, test_data)

Classification Model

Confusion Matrix

## Confusion Matrix and Statistics
## 
##             Actual
## Predictions  developed developing
##   developed        317         61
##   developing         6        859
##                                         
##                Accuracy : 0.9461        
##                  95% CI : (0.932, 0.958)
##     No Information Rate : 0.7401        
##     P-Value [Acc > NIR] : < 2.2e-16     
##                                         
##                   Kappa : 0.8672        
##                                         
##  Mcnemar's Test P-Value : 4.191e-11     
##                                         
##             Sensitivity : 0.9814        
##             Specificity : 0.9337        
##          Pos Pred Value : 0.8386        
##          Neg Pred Value : 0.9931        
##              Prevalence : 0.2599        
##          Detection Rate : 0.2550        
##    Detection Prevalence : 0.3041        
##       Balanced Accuracy : 0.9576        
##                                         
##        'Positive' Class : developed     
## 

Classification Model

ROC Curve

Conclusion

In this project,

• Analyzed electricity demand data for two groups of countries in the last 30 years

• Explored different data visualization techniques, including bar plots, line plots, and violin plots, to understand the distribution and trends in the data.

• Used statistical analysis techniques such as hypothesis testing, t-tests, and ANOVA to explore the relationships between different variables and draw inference from the data.

Overall, We found that electricity demand increased over time and that developed countries had statistically higher demand than developing countries. Our findings may have important insights for policymakers and energy companies looking to understand global trends in electricity consumption and plan for the future.