Skip to contents

Plot Gallery

Source: vignettes/articles/plot-gallery.Rmd
plot-gallery.Rmd

This gallery showcases every visualization function in ggExametrika. Each plot is generated from real model output using the exametrika package. All functions return standard ggplot objects, so you can further customize them with + theme(), + labs(), etc.

Setup

Model Fitting

We fit all models up front.

# --- IRT (binary, 15 items, 500 students) ---
result_irt <- IRT(J15S500, model = 2)

# --- GRM (ordinal, 5 items, 1000 students) ---
result_grm <- GRM(J5S1000)

# --- LCA (binary, 15 items, 500 students, 3 classes) ---
result_lca <- LCA(J15S500, ncls = 3)

# --- LRA (binary, 15 items, 500 students, 4 ranks) ---
result_lra <- LRA(J15S500, nrank = 4)

# --- LRAordinal (ordinal, 15 items, 3810 students, 4 ranks) ---
result_lra_ord <- LRA(J15S3810, nrank = 4, dataType = "ordinal")

# --- Biclustering: binary (35 items, 515 students) ---
result_bic <- Biclustering(J35S515, nfld = 5, nrank = 6)

# --- Biclustering: ordinal (35 items, 500 students, 5 categories) ---
result_ord_bic <- Biclustering(J35S500, ncls = 5, nfld = 5)

# --- Biclustering: nominal (20 items, 600 students, 4 categories) ---
result_nom_bic <- Biclustering(J20S600, ncls = 5, nfld = 4)

Network models (BNM, LDLRA, LDB, BINET) require explicit graph structure input (an igraph DAG or edge CSV file).

# --- BNM (5 items, 10 students, simple DAG) ---
bnm_dag <- igraph::make_empty_graph(n = 5, directed = TRUE)
igraph::V(bnm_dag)$name <- J5S10$ItemLabel
bnm_dag <- igraph::add_edges(bnm_dag, c(1, 3, 2, 4, 3, 5))
result_bnm <- BNM(J5S10, g = bnm_dag)

# --- LDLRA (12 items, 5000 students, 3 ranks, same DAG per rank) ---
ldlra_g <- igraph::graph_from_data_frame(
  data.frame(
    from = c("Item01", "Item02", "Item03"),
    to = c("Item02", "Item03", "Item04")
  ),
  directed = TRUE
)
result_ldlra <- LDLRA(J12S5000,
  ncls = 3,
  g = list(ldlra_g, ldlra_g, ldlra_g)
)

# --- LDB (35 items, 515 students, 3 fields, 3 ranks) ---
ldb_conf <- rep(1:3, length.out = 35)
ldb_g <- igraph::graph_from_data_frame(
  data.frame(
    from = c("Field01", "Field02"),
    to = c("Field02", "Field03")
  ),
  directed = TRUE
)
result_ldb <- LDB(J35S515,
  ncls = 3, conf = ldb_conf,
  g_list = list(ldb_g, ldb_g, ldb_g)
)

# --- BINET (35 items, 515 students, 3 classes, 3 fields) ---
binet_edges <- data.frame(
  From  = c(1, 2, 1, 2, 1, 2),
  To    = c(2, 3, 2, 3, 2, 3),
  Field = c(1, 1, 2, 2, 3, 3)
)
binet_edge_file <- tempfile(fileext = ".csv")
write.csv(binet_edges, binet_edge_file, row.names = FALSE)
result_binet <- BINET(J35S515,
  ncls = 3, nfld = 3,
  conf = ldb_conf, adj_file = binet_edge_file,
  verbose = FALSE
)
unlink(binet_edge_file)

1. IRT Models

Item Response Theory (2PL, 3PL, 4PL) visualization for binary response data. Data: J15S500 (15 items, 500 students).

plotICC_gg — Item Characteristic Curve

Probability of a correct response as a function of ability (θ\theta). Returns a list of plots (one per item).

icc_plots <- plotICC_gg(result_irt)
icc_plots[[1]]

combinePlots_gg(icc_plots, selectPlots = 1:6)

plotICC_overlay_gg — ICC Overlay

All item curves overlaid on a single plot for easy comparison.

plotICC_overlay_gg(result_irt)

plotIIC_gg — Item Information Curve

How precisely each item measures ability at different θ\theta levels.

iic_plots <- plotIIC_gg(result_irt)
iic_plots[[1]]

combinePlots_gg(iic_plots, selectPlots = 1:6)

plotIIC_overlay_gg — IIC Overlay

All item information curves on a single plot.

plotIIC_overlay_gg(result_irt)

plotTIC_gg — Test Information Curve

Total test information (sum of all item information functions).

plotTIC_gg(result_irt)

plotTRF_gg — Test Response Function

Expected total score as a function of ability.

plotTRF_gg(result_irt)

2. GRM (Graded Response Model)

Visualization for ordered polytomous response data. Data: J5S1000 (5 items, 1000 students).

plotICRF_gg — Item Category Response Function

Probability of each response category as a function of ability. Returns a list of plots (one per item).

icrf_plots <- plotICRF_gg(result_grm)
icrf_plots[[1]]

combinePlots_gg(icrf_plots, selectPlots = 1:5)

Note: plotIIC_gg() and plotTIC_gg() also accept GRM output. See the Getting Started vignette for examples.

3. Latent Class / Rank Analysis

Visualization for discrete latent variable models. Data: J15S500 with LCA (3 classes) and LRA (4 ranks).

plotIRP_gg — Item Reference Profile

Probability of a correct response for each item within each latent class or rank. Returns a list of plots (one per item).

irp_plots <- plotIRP_gg(result_lca)
irp_plots[[1]]

combinePlots_gg(irp_plots, selectPlots = 1:6)

plotFRP_gg — Field Reference Profile

Correct response rate by field for each class/rank. Returns a list of plots (one per field).

Note: plotFRP_gg() requires a model with field structure (Biclustering, IRM, LDB, or BINET). Shown here with binary Biclustering on J35S515 (5 fields, 6 ranks).

plotFRP_gg(result_bic)

plotTRP_gg — Test Reference Profile

Number of students and expected test score per class/rank.

plotTRP_gg(result_lca)

plotLCD_gg — Latent Class Distribution

Class membership distribution (LCA).

plotLCD_gg(result_lca)

plotLRD_gg — Latent Rank Distribution

Rank membership distribution (LRA).

plotLRD_gg(result_lra)

plotCMP_gg — Class Membership Profile

Membership probability profile for each student across all classes. Returns a list of plots (one per student).

cmp_plots <- plotCMP_gg(result_lca)
combinePlots_gg(cmp_plots, selectPlots = 1:6)

plotRMP_gg — Rank Membership Profile

Membership probability profile for each student across all ranks. Returns a list of plots (one per student).

rmp_plots <- plotRMP_gg(result_lra)
combinePlots_gg(rmp_plots, selectPlots = 1:6)

4. Biclustering

Biclustering simultaneously clusters items (into fields) and students (into ranks/classes). This section covers binary, ordinal, and nominal Biclustering visualizations.

Binary Biclustering

Data: J35S515 (35 items, 515 students, 5 fields, 6 ranks).

plotCRV_gg — Class Reference Vector

All class profiles in a single plot with one line per class.

plotCRV_gg(result_bic)

plotRRV_gg — Rank Reference Vector

All rank profiles in a single plot with one line per rank.

plotRRV_gg(result_bic)

plotArray_gg — Array Plot

Visualizes the data matrix as a heatmap showing the block-diagonal structure after biclustering.

plotArray_gg(result_bic)

#> TableGrob (1 x 2) "arrange": 2 grobs
#>   z     cells    name           grob
#> 1 1 (1-1,1-1) arrange gtable[layout]
#> 2 2 (1-1,2-2) arrange gtable[layout]

Clustered array only:

plotArray_gg(result_bic, Original = FALSE, Clusterd = TRUE)
#> [[1]]

Ordinal Biclustering

Data: J35S500 (35 items, 500 students, 5 categories, 5 fields, 5 classes).

plotFCRP_gg — Field Category Response Profile

Category probability plot with two display styles.

Line style (default):

plotFCRP_gg(result_ord_bic, style = "line")

Bar style:

plotFCRP_gg(result_ord_bic, style = "bar")

plotFCBR_gg — Field Cumulative Boundary Reference

Boundary probabilities P(Q≥k)P(Q \geq k) for each field across latent classes. Ordinal Biclustering only.

plotFCBR_gg(result_ord_bic)

Select specific fields:

plotFCBR_gg(result_ord_bic, fields = 1:3)

plotScoreField_gg — Expected Score Heatmap

Expected score for each field at each class/rank, displayed as a heatmap.

plotScoreField_gg(result_ord_bic)

Nominal Biclustering

Data: J20S600 (20 items, 600 students, 4 categories, 4 fields, 5 classes).

plotFCRP_gg (nominal) 

plotFCRP_gg(result_nom_bic, style = "line")

plotScoreField_gg (nominal) 

plotScoreField_gg(result_nom_bic)

Polytomous stat parameter

plotFRP_gg(), plotCRV_gg(), and plotRRV_gg() accept a stat parameter for polytomous data: "mean" (default), "median", or "mode".

p_mean <- plotFRP_gg(result_ord_bic, stat = "mean") +
  ggplot2::ggtitle("stat = 'mean'")
p_median <- plotFRP_gg(result_ord_bic, stat = "median") +
  ggplot2::ggtitle("stat = 'median'")
p_mode <- plotFRP_gg(result_ord_bic, stat = "mode") +
  ggplot2::ggtitle("stat = 'mode'")
gridExtra::grid.arrange(p_mean, p_median, p_mode, ncol = 2)

5. LRAordinal / LRArated

Polytomous Latent Rank Analysis with specialized score-level visualizations. Data: J15S3810 (15 items, 3810 students, 4 ranks, ordinal).

plotScoreFreq_gg — Score Frequency Distribution

Density distribution of scores with vertical lines at rank boundary thresholds.

plotScoreFreq_gg(result_lra_ord)

plotScoreRank_gg — Score-Rank Heatmap

Joint distribution of observed scores and estimated ranks. Darker cells indicate higher frequency.

plotScoreRank_gg(result_lra_ord)

plotICRP_gg — Item Category Reference Profile

Probability of selecting each response category across latent ranks. Probabilities at each rank sum to 1.0.

plotICRP_gg(result_lra_ord, items = 1:4)

plotICRP_gg(result_lra_ord, items = 5:8)

plotICBR_gg — Item Category Boundary Response

Cumulative probability curves P(response≥k)P(\text{response} \geq k) for each category boundary. LRAordinal only.

plotICBR_gg(result_lra_ord, items = 1:4)

6. Network Models (DAG)

Directed Acyclic Graph and network model visualizations. All four DAG models are supported by plotGraph_gg(): BNM (items), LDLRA (items per rank), LDB (fields per rank), and BINET (classes + fields integrated).

plotGraph_gg — DAG Visualization

BNM (Bayesian Network Model)

The simplest DAG: item-to-item dependency structure. Data: J5S10 (5 items, 10 students).

dag_bnm <- plotGraph_gg(result_bnm)
dag_bnm[[1]]

Different layout directions:

bnm_dirs <- lapply(c("BT", "TB", "LR", "RL"), function(d) {
  plotGraph_gg(result_bnm, direction = d, title = paste("BNM -", d))[[1]]
})
gridExtra::grid.arrange(grobs = bnm_dirs, ncol = 2)

LDLRA (Locally Dependent Latent Rank Analysis)

One DAG per rank, showing how item dependencies change across ranks. Data: J12S5000 (12 items, 5000 students, 3 ranks).

dag_ldlra <- plotGraph_gg(result_ldlra)
combinePlots_gg(dag_ldlra)

LDB (Locally Dependent Biclustering)

One DAG per rank with field-level nodes (green diamonds). Data: J35S515 (35 items, 515 students, 3 fields, 3 ranks).

dag_ldb <- plotGraph_gg(result_ldb)
combinePlots_gg(dag_ldb)

BINET (Bicluster Network Model)

Integrated network with two node types: Class nodes (blue squares) and Field nodes (green diamonds) as intermediates. Data: J35S515 (35 items, 515 students, 3 classes, 3 fields).

dag_binet <- plotGraph_gg(result_binet)
dag_binet[[1]]

With legend showing node types:

plotGraph_gg(result_binet,
  show_legend = TRUE,
  legend_position = "right", title = "BINET with Legend"
)[[1]]

Comparing all four DAG models 

p_bnm <- plotGraph_gg(result_bnm, title = "BNM (Items)")[[1]]
p_ldlra <- plotGraph_gg(result_ldlra, title = "LDLRA (Items)")[[1]]
p_ldb <- plotGraph_gg(result_ldb, title = "LDB (Fields)")[[1]]
p_binet <- plotGraph_gg(result_binet,
  title = "BINET (Class+Field)",
  show_legend = TRUE
)[[1]]
gridExtra::grid.arrange(p_bnm, p_ldlra, p_ldb, p_binet, ncol = 2)

plotFieldPIRP_gg — Field Parent Item Reference Profile

Shows how field performance varies based on parent field scores, for each rank. LDB only. Returns a list of plots (one per rank).

fpirp_plots <- plotFieldPIRP_gg(result_ldb)
combinePlots_gg(fpirp_plots)

plotLDPSR_gg — Latent Dependence Passing Student Rate

Item-level correct response rate comparing parent and child classes at each DAG edge. BINET only. Returns a list of plots (one per edge).

ldpsr_plots <- plotLDPSR_gg(result_binet)
ldpsr_plots[[1]]

combinePlots_gg(ldpsr_plots, selectPlots = 1:min(6, length(ldpsr_plots)))

7. Common Options Demo

All plot functions share a consistent set of customization options. Every function returns a ggplot object, so you can add further layers with standard ggplot2 syntax.

Parameter Type Default Description
title logical or character TRUE TRUE = auto title, FALSE = hidden, character = custom
colors character vector NULL NULL = colorblind-friendly palette, or custom colors
linetype character or numeric varies "solid", "dashed", "dotted", etc.
show_legend logical varies Show or hide the legend
legend_position character "right" "right", "top", "bottom", "left", "none"

title 

p1 <- plotICC_overlay_gg(result_irt, items = 1:5, title = TRUE)
p2 <- plotICC_overlay_gg(result_irt, items = 1:5, title = FALSE)
p3 <- plotICC_overlay_gg(result_irt, items = 1:5, title = "My Custom Title")
gridExtra::grid.arrange(p1, p2, p3, ncol = 3)

colors 

plotICRF_gg(result_grm,
  items = 1,
  colors = c("#D81B60", "#1E88E5", "#FFC107", "#004D40", "#7B1FA2")
)
#> [[1]]

plotCRV_gg(result_bic,
  colors = c(
    "steelblue", "coral", "forestgreen",
    "mediumpurple", "goldenrod", "hotpink"
  )
)

linetype 

p1 <- plotTIC_gg(result_irt, linetype = "solid", title = "solid")
p2 <- plotTIC_gg(result_irt, linetype = "dashed", title = "dashed")
p3 <- plotTIC_gg(result_irt, linetype = "dotdash", title = "dotdash")
gridExtra::grid.arrange(p1, p2, p3, ncol = 3)

show_legend and legend_position 

p1 <- plotICC_overlay_gg(result_irt,
  items = 1:5,
  show_legend = TRUE, legend_position = "right", title = "right"
)
p2 <- plotICC_overlay_gg(result_irt,
  items = 1:5,
  show_legend = TRUE, legend_position = "bottom", title = "bottom"
)
p3 <- plotICC_overlay_gg(result_irt,
  items = 1:5,
  show_legend = TRUE, legend_position = "top", title = "top"
)
p4 <- plotICC_overlay_gg(result_irt,
  items = 1:5,
  show_legend = FALSE, title = "show_legend = FALSE"
)
gridExtra::grid.arrange(p1, p2, p3, p4, ncol = 2)

Combining options 

plotICRF_gg(result_grm,
  items = 1,
  title = "Fully Customized ICRF",
  colors = c("#E41A1C", "#377EB8", "#4DAF4A", "#984EA3", "#FF7F00"),
  linetype = "solid",
  show_legend = TRUE,
  legend_position = "bottom"
)
#> [[1]]

Post-hoc ggplot2 customization

Since all functions return ggplot objects, you can add any ggplot2 layer:

plotTIC_gg(result_irt) +
  ggplot2::theme_minimal() +
  ggplot2::labs(
    subtitle = "Custom subtitle via ggplot2",
    caption = "Generated with ggExametrika"
  )