The AdaptiveConformal
package can be installed using the remotes
package: ::: {.cell}
Hide/Show the code
::install_github("herbps10/AdaptiveConformal") remotes
From e02852d86350a120ea0fe95b2dd5a9a7099db0d4 Mon Sep 17 00:00:00 2001
From: Quarto GHA Workflow Runner 3 Algorithms
For demonstration purposes we assume we have access to unbiased predictions \hat{\mu}_t = 0 for all t \in \llbracket T \rrbracket. Throughout we set the target empirical coverage to \alpha = 0.8.
AdaptiveConformal
R
packageThe ACI algorithms described in the previous section have been implemented in the open-source and publically available R
package AdaptiveConformal
, available at https://github.com/herbps10/AdaptiveConformal. CIn this section, we briefly introduce the main functionality of the package. Comprehensive documentation is, including several example vignettes, is included with the package.
The AdaptiveConformal
package can be installed using the remotes
package:
The AdaptiveConformal
package can be installed using the remotes
package: ::: {.cell}
::install_github("herbps10/AdaptiveConformal") remotes
The ACI algorithms are accessed through the aci
function, which takes as input a vector of observations (y_t) and a vector or matrix of predictions (\hat{y}_t). Using the data generating process from the running example to illustrate, we can fit the original ACI algorithm with learning rate \gamma = 0.1:
:::
+The ACI algorithms are accessed through the aci
function, which takes as input a vector of observations (y_t) and a vector or matrix of predictions (\hat{y}_t). Using the data generating process from the running example to illustrate, we can fit the original ACI algorithm with learning rate \gamma = 0.1: ::: {.cell}
set.seed(532)
<- running_example_data(N = 5e2)
data <- aci(data$y, data$yhat, alpha = 0.8, method = "ACI", parameters = list(gamma = 0.1)) fit
The available parameters for each method can be found in the documentation for the aci
method, accessible with the command ?aci
. The resulting conformal prediction intervals can then be plotted using the plot
function:
::: The available parameters for each method can be found in the documentation for the aci
method, accessible with the command ?aci
. The resulting conformal prediction intervals can then be plotted using the plot
function: ::: {.cell}
plot(fit)
AdaptiveC
The properties of the prediction intervals can also be examined using the summary
function:
::: The properties of the prediction intervals can also be examined using the summary
function: ::: {.cell}
summary(fit)
AdaptiveC
Mean interval width: 0.354
Mean interval loss: 0.498
:::
AdaptiveConformal
R
packageThe ACI algorithms described in the previous section have been implemented in the open-source and publically available R
package AdaptiveConformal
, available at https://github.com/herbps10/AdaptiveConformal. CIn this section, we briefly introduce the main functionality of the package. Comprehensive documentation is, including several example vignettes, is included with the package.
The AdaptiveConformal
package can be installed using the remotes
package:
The AdaptiveConformal
package can be installed using the remotes
package: ::: {.cell}
::install_github("herbps10/AdaptiveConformal") remotes
The ACI algorithms are accessed through the aci
function, which takes as input a vector of observations (y_t) and a vector or matrix of predictions (\hat{y}_t). Using the data generating process from the running example to illustrate, we can fit the original ACI algorithm with learning rate \gamma = 0.1:
:::
+The ACI algorithms are accessed through the aci
function, which takes as input a vector of observations (y_t) and a vector or matrix of predictions (\hat{y}_t). Using the data generating process from the running example to illustrate, we can fit the original ACI algorithm with learning rate \gamma = 0.1: ::: {.cell}
set.seed(532)
<- running_example_data(N = 5e2)
data <- aci(data$y, data$yhat, alpha = 0.8, method = "ACI", parameters = list(gamma = 0.1)) fit
The available parameters for each method can be found in the documentation for the aci
method, accessible with the command ?aci
. The resulting conformal prediction intervals can then be plotted using the plot
function:
::: The available parameters for each method can be found in the documentation for the aci
method, accessible with the command ?aci
. The resulting conformal prediction intervals can then be plotted using the plot
function: ::: {.cell}
plot(fit)
AdaptiveC
The properties of the prediction intervals can also be examined using the summary
function:
::: The properties of the prediction intervals can also be examined using the summary
function: ::: {.cell}
summary(fit)
AdaptiveC
Mean interval width: 0.354
Mean interval loss: 0.498
:::
d)nPVKM*qe;4YLJ8
z1T&dTDOSqr3^*LM%uCn XOblCXga0hQ=+?BK{u1h{HW5B4!uZDw6>MT8tN{9E#6#?
z4l=___Qg@f0Vx}Zk6<;~4_xzwDWW^{uo$fnJRt(6Q(7+>NDmQay9&H$xwZYCvgwb9
zlMC0$2qjHYAf7$@RxTTe8QyA0W#$$nMO7xb#10cS^CQ9re3t~K`w=lJA}z%lM?G4y
zz?s%N4&1`D(NQVIfo0T!yaG4C!S0@eNMYEI_HZWM5ig>9*sM9kHvRb3$it0`Z?7MH
zc>-k I~l4C+;dPfWBz4%apVij>mplQ!Ox
z&XtvlpJ1e$A*4+(CUvn-bC9XAIr@pqu;|YbY||X!h(+jZ6}Fv$W`_@5VFt8v{N%>S)j>MRaBTL!XP
zNpT<;ibIasj|Z$j{G5`n=p|DERn!aZ+y}a88gP?6M1ss9mBeOHm0B>JQ-+R>TeJD)
z1oEE19ZrNTKifWf@YO(rf>^6yu0BdldG#gj%9UngvwDkgmVwnc;~!y$;UkOG2`ki!
zy&wS!4Mc
vn+N7
zVP!0E9@K^~#xb3N6qRb0h;RgJ5Ir;uIknyrHb3+9%2KL#NpyYlW>W2CDQ~Ks*9*G;<^@Y6KHtVYVpzhRGwyx0={8tbLBFnPJT-oh=)EiY4f{;eVB*i4FQbujcQ8
z-YD$;haj84HfNBw8$q*gr-`qj+^g?abezn-hMHG=7vke
rQaIknC_`@KQ1xoWO
zvzl_hB|B{-GhoFz;4{+bVhDs!6Z@%WV|2x!;Oow5pg?K)A@H)T34DW_rVwgBe7BGY
z6x2GG!d#C|-c4z3E-io#hCN=7f`Xejw()8AOfaO3pd!nsz;}V+b2(AEgbl%=h9~&%R5%Du1xFD5%7Qqn|Q#!(&KF+
zj%HK;`*VXK{~DiHg-?9rTqAHzyr5-vhjOHXvccIH!QJs;rOIQ2eoLGQ$QlE~=9?QR
z6v9V6avw1pX>r@z*V)5k12ii!;5T?g6Wsj(-~Lwt(7RmDE8a_j9c6vAR9;OWzEKvK
z+83c|4y!cVZ=WH1C%