$\hat{y} =X{dc}b{dc}$ => continuous data
predict(m;type=:only_epoch) / predict(m::TimeContinuous;type=:continuous)
predict new data
$\hat{y} = X_{new}b$ => epoched and continuous data
predict(m,DataFrame(:continuous=[1,2],:categorical["A","B"]) => always no overlap
new stuff
residuals(m,data) = data .- predict(m)
"event-based" predict
$\hat{y}$ but you have case (2), but want results only around certain latencies $\pm$, so case (1)
predict(m::TimeContinuous,latencies,tau) & predict(m::TimeContinuous,evts::DataFrame,tau) & predict(m::TimeContinuous,:basisname/eventname)predict(m::TimeContinuous;type=:epoch) == predict.(m,:basisnames/eventnames)
=> full overlap
predicttable
predicttable => wrapper around predict, returns an effects-output-style DataFrame
partial overlap
predict(m,exclude=[:eventnameA]) => put all other coefficients to 0 and call predict(X,b) => some overlap
predict(m,include=[:eventnameA])
@ReneSkukies all yours :)
What do we want?
Climatejustice!
rename stuff
"standard" predict
predict(m;type=:only_epoch)
/predict(m::TimeContinuous;type=:continuous)
predict new data
$\hat{y} = X_{new}b$ => epoched and continuous data
predict(m,DataFrame(:continuous=[1,2],:categorical["A","B"])
=> always no overlapnew stuff
residuals(m,data) = data .- predict(m)
"event-based" predict
$\hat{y}$ but you have case (2), but want results only around certain latencies $\pm$, so case (1)
predict(m::TimeContinuous,latencies,tau)
&predict(m::TimeContinuous,evts::DataFrame,tau)
&predict(m::TimeContinuous,:basisname/eventname)
predict(m::TimeContinuous;type=:epoch)
==predict.(m,:basisnames/eventnames)
=> full overlap
predicttable
predicttable
=> wrapper around predict, returns an effects-output-style DataFramepartial overlap
predict(m,exclude=[:eventnameA])
=> put all other coefficients to 0 and callpredict(X,b)
=> some overlappredict(m,include=[:eventnameA])
Stimulus,Fixations,Fixation,Fixation
I want: