This paper aims to understand the dynamics of the spread of COVID-19 for Nepal . It is carried out with the help of multivariate statistics techniques . Direct relationships among variables are obvious, as they are easily seen and measured . But, hidden variables and their interrelationships also have a significant effect on the spread of a pandemic . Multinomial logistic regression, odds ratio, linear mixed-effect models, and principal component analysis are used here to analyze these hidden variables and their interrelationships . Also, such studies are very important for countries with limited and scarce data . These countries do not have a backbone of good-quality official records . Understanding the spread of a disease in a developing country also helps in management and eradication of that disease . The multivariate daily data of new cases, deaths, recovered, total cases, total deaths, total recovered, and total infected (isolated) are used here . The daily incidence of new cases is also modeled here using nonlinear regression . Two best nonlinear models are discussed here . ARIMA models are used for analyzing and forecasting the progression of the variables for two months into the future . The impact of government restriction in the form of strict lockdown 1, partially relaxed lockdown 1, completely relaxed lockdown 1, and strict lockdown 2 is minutely analyzed . These controls were exercised to curtail the spread of the pandemic . The role of these controls in curbing the spread of the pandemic is also studied here . The results obtained from this study can be applied to other countries of South Asia and Africa.