In this research, we are faced to a longitudinal study.
The central concern of longitudinal research, however, revolves around the description of patterns of stability and change, and the explanation of how and why change does or does not take place Overall conditions governing our problem oN participants are recruited. (Here, we have three methods that needed to be assessed) oAll participants are assessed on T occasions. oThe assessment schedule is identical for all participants. oThere are some missing data. Our Goal
The goal is to build a model to incorporate predictors of individual
variability in growth Why growth curve model • Growth curve model is used to model development of this technology explicitly as a function of time. • To capture continuous trajectory of change, will approach precisely same data structure from different perspective • Will build model for data that estimates change over time within each individual and then compare change across individuals Assumptions • All subjects in a given population are assumed to have developmental curves of the same functional form (e.g., all linear), but the parameters describing their curves may differ. In our work, we discuss cost and value of Hydrogen in next years. • The correlation between initial level and growth rate is zero. In other words, the initial level has no predictive value for the growth rate. Approaches It can be approached from different perspectives but in here we study two frameworks; • Multilevel Regression (MLR) • Latent Growth Curve Model (LGC). Approaches Both MLR and LGC incorporate the factor ‘time’ explicitly. Within the MLR framework time is modeled as an independent variable The LGC approach adopts a latent variable view; Time is incorporated as specific constrained values for the factor loadings of the latent variable that represents the slope of the growth curve; The mean and covariance structure of the latent variables in LGC analysis correspond to the fixed and random effects in MLR analysis, Describing equations & involved parameters The main equations for a repeatedly measured variable Ɣ𝑡𝑖 of individual i, at occasion t, may be written as:,
Ɣ𝑡𝑖 = ƛ0𝑡 ƞ0𝑖 +ƛ1𝑡 ƞ1𝑖 +Ɣ𝑡𝑖 𝑋𝑡𝑖 + Ɵ𝑡𝑖
ƞ0𝑖 = Ʋ0𝑖 + Ɣ0 𝑍𝑖 + ζ0𝑖 ƞ1𝑖 = Ʋ1 + Ɣ1 𝑍𝑖 + ζ1𝑖 where ƛ1𝑡 denotes the time of measurement and ƛ0𝑡 a constant equal to the value of 1. The individual intercept and slope of the growth curve are represented by ƞ0𝑖 and ƞ1𝑖 , respectively. Describing equations & involved parameters Ʋ0 and Ʋ1 are expectations, and random departures or residuals are ζ0𝑖 and ζ1𝑖 , respectively Ɣ2𝑡 represents the effect of the time-varying covariate 𝑋𝑡𝑖 ; Ɣ0 and Ɣ1 are the effects of the time-invariant covariate on the initial level and linear slope. The differences between these two frameworks refer to procedure of treatment with some parameters and the way is time is incorporated in the model. In MLR ƞ0𝑖 &ƞ1𝑖 are the random parameters but in LGC, these are latent variables and ƛ0𝑡 &ƛ1𝑡 are parameters. Describing equations & involved parameters MLR is essentially a univariate approach, with time points treated as observation of the same Variable But LGC is a multivariate approach with each Time point is treated as a separate Variable In Our Case This a growth analysis of power to gas technology, where individual data are collected for a sample of individuals at different occasions in their development. In fact, with this modelling we could summarize the behavior of existing data in an understandable format. We need time-adjacent predictions and growth model can estimate smoothed time trajectories within individuals. It can be conceptualized as repeated measures nested within individuals. We are curious to know, what is the starting point of this pool of data and where this trajectory will go over the time or the slope of obtained profile. In Our Case This modelling will average over the existing data during determined time. There are different methods of generating hydrogen in the context of power to gas technology. In the specified time period we can find an average of whole cost related to the generation of hydrogen. It can show inter individual differences in each trajectory and intra individual changes among the trajectories. Regarding the mentioned points we choose MLR framework for our studies.