**4.3 Forecasting results**

For prediction, we used a Weibull based Long-Short-Term-Memory approach (W-LSTM) [34]. According to the author of W-LSTM, the model outperformed ARIMA and other LSTM variants. Moreover, the network got 82% of accuracy. In **Figures 9**–**12**, we show the predictions of total confirmed cases and daily cases for

**Figure 11.**

*Prediciton of total and daily confirmed cases in Russia, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

**Figure 12.**

*Prediciton of total and daily confirmed cases in France, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

**Figure 13.**

*Prediciton of total and daily confirmed cases in Argentina, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

US, India, Russia and France. Additionally, in **Figures 13**–**17**, we present the predictions for Latin American countries.

In **Figure 18a**, we plotted the total confirmed cases predictions for US, India, Russia, France and Brazil, using MTES algorithm. Additionally, In **Figure 18b**, we plotted the total confirmed cases predictions for Peru, Argentina, Colombia, Chile and Mexico. We know that, MTES is usually well used for short time series

**Figure 14.**

*Prediciton of total and daily confirmed cases in Chile, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

**Figure 15.**

*Prediciton of total and daily confirmed cases in Colombia, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

**Figure 16.**

*Prediciton of total and daily confirmed cases in Mexico, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

*COVID-19 Pandemic: Analysis and Statistics of Confirmed Cases DOI: http://dx.doi.org/10.5772/intechopen.98891*

**Figure 17.**

*Prediciton of total and daily confirmed cases in Peru, using LSTM. (a) Total confirmed cases. (b) Daily confirmed cases.*

#### **Figure 18.**

*Prediciton of total and daily confirmed cases in Peru, using LSTM. (a) Predicitons for US, India, Russia, France and Brazil. (b) Predictions for Peru, Argentinca, Colombia, Chile and Mexico.*

prediction. For that reason, the India confirmed cases predictions shows an increased trend, due to the increasing behavior during the last weeks.

#### **5. Conclusions**

The coronavirus COVID-19 pandemic caused strain on all the world getting abundant deaths and forcing lock downs to contain the spread. However, the scientific community was not left behind because it was developed a lot of projects like vaccines candidates, analysis of the confirmed cases, and forecast of confirmed cases and deaths.

The behavior and evolution of confirmed cases is different for each country. Moreover, there are several factors that increase or mitigate the COVID-19 evolution like: population, health system, social behavior and the overestimation of some authorities. Moreover, in order to evaluate the impact of the pandemic, we need to evaluate the number of confirmed cases, population, deaths, etc. For instance, despite US has the major number of confirmed cases, it has a low death rate of 1.77%.

The death rate, is a good metric to evaluate the impact of COVID-19 over population. For example, we noticed that Mexico has the highest death rate in this study (8.85%). After review, we found out, that the reason of this high death rate, is the percentage of over weighted people in Mexico (40.2% for males and 36.1% for females). According to researches, obesity is considered the main risk factor of death by COVID-19.

Additionally, we reviewed the variants and vaccination projects for COVID-19. Thankfully, we only have VOI and VOC variants. Furthermore, there are several vaccination projects around the world. Some countries, like US has started a massive vaccination plan, as a consequence, the number of confirmed cases and deaths, show a decreasing behavior.

Finally, we made some predictions. We used W-LSTM and MTES to predict the total and daily confirmed cases in US, India, Russia, France, Argentina, Colombia, Chile, Peru and Mexico. According to the results, W-LSTM showed a more realistic prediction than MTES.
