Selected article for: "accurate model and logistic model"

Author: Svetoslav Bliznashki
Title: A Bayesian Logistic Growth Model for the Spread of COVID-19 in New York
  • Document date: 2020_4_7
  • ID: lhv83zac_18
    Snippet: Instead, we attempted to construct a more accurate model within the same logistic growth paradigm in a different way: we introduced weights to our data points with later data points receiving higher weights than older ones in the hope that this will alleviate some of the problems observed above. Specifically, we weighted the points according to a Rectified Linear-like function (e.g. Glorot et al., 2011) whereby the first 20 observations received .....
    Document: Instead, we attempted to construct a more accurate model within the same logistic growth paradigm in a different way: we introduced weights to our data points with later data points receiving higher weights than older ones in the hope that this will alleviate some of the problems observed above. Specifically, we weighted the points according to a Rectified Linear-like function (e.g. Glorot et al., 2011) whereby the first 20 observations received constant (0.008) low weights and the last 8 observations received linearly increasing higher weights (last 8 weights=[0.77 1.55 2.32 3.09 3.87 4.64 5.41 6.19]). The idea behind this scheme was to try to force the model to account better for the observations following the approximately linear trend observed in the upper half of Fig. 2 . Note also that the weights sum to the number of original observations (28). The weights pattern is shown in Fig. 3 . In the subsequent simulations we used the proposed weights in order to weigh the likelihood function of the model. Following Simeckova (2005) , assuming we have observations Y 1 , …Y n and Y i has density f i (Y i |θ) where θ is the vector of parameters (see Eq.

    Search related documents:
    Co phrase search for related documents
    • accurate model and Eq parameter: 1
    • better account and data point: 1
    • data point and high weight: 1
    • data point and likelihood function: 1, 2
    • data point and low weight: 1
    • high weight and low weight: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25
    • high weight increase and low weight: 1, 2