WanjohiWanjohi/corona-spread-prediction

A repository that focuses on epidemiology modelling , with SEIR in particular as applied to the prediction of the spread COVID-19 virus in 2020 for the week of 10th August in Senegal ## Problem Statement In late 2019 , the spread of the corona virus was limited to the Asias and only became a matter of gloabla precedence in February 2020. The spread of the disease has been managed by the imposition of different measures by different governments towards maintaining social distancing. In the implementing of these measures, it is important to find out which ones bear the most result in slowing down the infection rate and which measures could be done away with a change point analysis to quantify the impact of policy interventions to slow the spread of COVID-19 ## Objective For the week of 10th August , it is the goal of this project to find the possible number of infected persons given the different policies implemented by the Senegalese government. The change points considered prior are : - 10/03/2020 : All learning and religious instituions shut down - 16/03/2020: Travel ban declared both locally and externally - 31/03/2020: Declaration of a national state of emergency ## Tools and Methods used 1. The SIR model uses the Susceptible-Infected-Recovered rate where the susceptible population reduces as the number of those infected increase ( an inverse relationship). There are four parameters to be estimated in the SIR model from data: - transmission rate β - diagnosis rate ρ - the initial population size I0 when time (t) = 0 (t=0) - variance q=1/τ for the noise distribution in the data. More information about the SIR model can be found in this paper by John Dehning et all (2020) here: science.sciencemag.org /6500/eabb9789.full.pdf 2. PyMC3 - a standard Bayseian modelling package in Python. More information about the PyMC3 modelcan be found here : docs.pymc.io 3. Markov Chain Monte Carlo Sampling : PyMC3 samples in multiple chains, or independent processes.In this case we use it as a simulation technique that can be used to find the posterior distribution and to sample from …