Interactive Data Visualization

A Parallel Guide in R and Python

1 Introduction

While static graphics are ideal for publication, interactive visualizations are essential for data exploration and quality assurance (QA). Interactive tooltips allow researchers to inspect individual outliers, verify data flags, and examine subject metadata without cluttering the plot canvas.

In this module, we demonstrate how to build interactive figures: * In R, we use ggiraph, which extends ggplot2 syntax by converting standard geoms into interactive HTML elements. * In Python, we use plotly (plotly.express and plotly.graph_objects), which creates web-ready JavaScript widgets from pandas DataFrames.

2 Interactive Postural Sway Across Conditions

In this first example, we make our multi-condition postural sway trajectories interactive. Hovering over any point displays subject metadata (Subject ID, Age, Gender, and Total Path Length), and hovering over an individual point highlights the subject’s entire trajectory across conditions.

R (ggplot2 & ggiraph)
Python (plotly)

Show Code

library(tidyverse)
library(ggiraph)

# 1. Load Summary Data & Aggregate to Subject Averages
DATA <- readRDS("./data/BDS/output/BDS_trial_summary.rds") %>% 
  mutate(
    Subject  = factor(Subject),
    AgeGroup = fct_relevel(AgeGroup, "Young", "Old")
  )

AVE_DATA <- DATA %>% 
  select(-Trial) %>%
  group_by(Subject, Vision, Surface) %>%
  summarise(
    across(where(is.numeric), ~ mean(.x, na.rm = TRUE)),
    across(!where(is.numeric), dplyr::first),
    .groups = "drop"
  ) %>%
  mutate(
    Condition = factor(paste0(Surface, ":", Vision),
                       levels = c("Firm:Open", "Firm:Closed", "Foam:Open", "Foam:Closed"))
  )

# 2. Build Tooltip Text
AVE_DATA <- AVE_DATA %>%
  mutate(tooltip_text = paste0(
    "<b>Subject ID:</b> ", Subject, "<br>",
    "<b>Age:</b> ", round(Age, 1), " years<br>",
    "<b>Gender:</b> ", Gender, "<br>",
    "<b>Sway (Path Length):</b> ", round(PathLength_Total, 1), " cm"
  ))

# 3. Define Colorblind Palette
cbPalette <- c("#1E6654", "#CC5500", "#954535", "#5B7C99")

# 4. Create Interactive ggplot
p_sway <- ggplot(data = AVE_DATA, 
                 aes(x = Condition, y = PathLength_Total,
                     tooltip = tooltip_text, data_id = Subject)) +
  geom_line(aes(group = Subject),
            color = "gray75", alpha = 0.5, linewidth = 0.5,
            position = position_dodge(width = 0.2)) +
  geom_point_interactive(aes(shape = Condition, col = Condition, group = Subject), 
                         size = 2.5, alpha = 0.7, position = position_dodge(width = 0.2),
                         hover_nearest = TRUE) +
  facet_wrap(~ AgeGroup) + 
  scale_color_manual(values = cbPalette) + 
  scale_fill_manual(values = cbPalette) +
  theme_bw() +
  labs(
    title = "Postural Sway: Between- and Within-Subject Effects",
    subtitle = "Hover points to inspect subject metadata; lines connect individual trajectories",
    x = NULL,
    y = "Total Path Length (cm)",
    shape = "Condition", col = "Condition", fill = "Condition"
  ) +
  theme(
    plot.title = element_text(size = 14, face = "bold"),
    plot.subtitle = element_text(size = 11, face = "italic", color = "gray30"),
    axis.title = element_text(size = 12, face = "bold"),
    axis.text.x = element_blank(),
    axis.text.y = element_text(size = 10),
    legend.position = "bottom",
    strip.text = element_text(size = 11, face = "bold")
  )

# 5. Render Interactive girafe Widget
girafe(
  ggobj = p_sway,
  options = list(
    opts_hover(css = "fill:red;stroke:black;r:5pt;"),
    opts_hover_inv(css = "opacity:0.2;"),
    opts_tooltip(css = "background-color:black;color:white;padding:8px;border-radius:4px;font-family:sans-serif;")
  )
)

Show Code

import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go

# 1. Load Summary Data & Aggregate to Subject Averages
DATA = pd.read_csv("./data/BDS/output/BDS_trial_summary.csv")
DATA["AgeGroup"] = pd.Categorical(DATA["AgeGroup"], categories=["Young", "Old"], ordered=True)

meta_cols = ["Subject", "Vision", "Surface", "AgeGroup", "Gender"]
AVE_DATA = DATA.groupby(meta_cols, as_index=False, observed=True).mean(numeric_only=True)
AVE_DATA["Condition"] = AVE_DATA["Surface"].astype(str) + ":" + AVE_DATA["Vision"].astype(str)

cond_order = ["Firm:Open", "Firm:Closed", "Foam:Open", "Foam:Closed"]
cbPalette = ["#1E6654", "#CC5500", "#954535", "#5B7C99"]
cond_colors = {cond: cbPalette[i] for i, cond in enumerate(cond_order)}

# 2. Build Base Scatter Plot
fig1 = px.scatter(
    AVE_DATA,
    x="Condition",
    y="PathLength_Total",
    color="Condition",
    symbol="Condition",
    facet_col="AgeGroup",
    category_orders={"AgeGroup": ["Young", "Old"], "Condition": cond_order},
    color_discrete_map=cond_colors,
    custom_data=["Subject", "Age", "Gender", "PathLength_Total"],
    labels={"PathLength_Total": "Total Path Length (cm)", "Condition": "Condition"},
    title="<b>Postural Sway: Between- and Within-Subject Effects</b><br><sup>Hover points to inspect subject metadata; lines connect individual trajectories</sup>"
)

# 3. Add Within-Subject Trajectory Lines Across Subplots
for i, age_grp in enumerate(["Young", "Old"], start=1):
    sub_df = AVE_DATA[AVE_DATA["AgeGroup"] == age_grp]
    for sub_id, grp in sub_df.groupby("Subject"):
        grp_sorted = grp.set_index("Condition").reindex(cond_order).dropna().reset_index()
        fig1.add_trace(
            go.Scatter(
                x=grp_sorted["Condition"],
                y=grp_sorted["PathLength_Total"],
                mode="lines",
                line=dict(color="rgba(180, 180, 180, 0.5)", width=1),
                showlegend=False,
                hoverinfo="skip"
            ),
            row=1, col=i
        )

# Move trajectory lines behind scatter points
n_conditions = len(cond_order)
fig1.data = tuple(list(fig1.data[n_conditions:]) + list(fig1.data[:n_conditions]))

# 4. Format Custom Hover Tooltip & Layout
fig1.update_traces(
    selector=dict(mode="markers"),
    marker=dict(size=8, opacity=0.8),
    hovertemplate=(
        "<b>Subject ID:</b> %{customdata[0]}<br>"
        "<b>Age:</b> %{customdata[1]:.1f} years<br>"
        "<b>Gender:</b> %{customdata[2]}<br>"
        "<b>Sway (Path Length):</b> %{customdata[3]:.1f} cm<extra></extra>"
    )
)

fig1.update_layout(
    template="plotly_white",
    legend=dict(orientation="h", y=-0.2, x=0.5, xanchor="center"),
    height=500
)

# Single final expression evaluated
fig1

3 Interactive Quality Assurance (Force Plate Calibration & Unweighting)

In this second example, we evaluate force plate calibration and screen for potential artifacts (such as a participant stepping off or unweighting the platform).

Under static gravity, vertical force $F_z$ should approximate body weight in Newtons:

\[F_z \approx \text{Mass (kg)} \times 9.81 \text{ m/s}^2\]

Hovering over flagged data points instantly reveals trial parameters, recorded body weight, and minimum vs. median $F_z$ values.

R (ggplot2 & ggiraph)
Python (plotly)

Show Code

# 1. Prepare Interactive Tooltips with Detailed QA Information
QA_data <- DATA %>%
  mutate(
    QA_tooltip = paste0(
      "<b>Subject:</b> ", Subject, " (", AgeGroup, ", ", Gender, ")<br>",
      "<b>Trial:</b> ", Trial, "<br>",
      "<b>Condition:</b> ", Surface, " - ", Vision, "<br>",
      "<b>Recorded Weight:</b> ", Weight, " kg<br>",
      "<b>Median Fz:</b> ", round(Fz_median, 1), " N<br>",
      "<b>Min Fz:</b> ", round(Fz_min, 1), " N<br>",
      "<b>Duration:</b> ", duration_s, "s (", n_samples, " samples)<br>",
      "<b>Flagged?:</b> ", ifelse(flag_any, "⚠️ YES", "✅ No")
    )
  )

# 2. Build Interactive Scatter Plot
qa_plot <- ggplot(QA_data, aes(x = Weight, y = Fz_median, tooltip = QA_tooltip, data_id = Trial)) +
  geom_abline(slope = 9.81, intercept = 0, linetype = "dashed", color = "gray50") +
  geom_point_interactive(aes(color = flag_any), size = 2.5, alpha = 0.7) +
  scale_color_manual(
    values = c("FALSE" = "#1E6654", "TRUE" = "#CC5500"), 
    labels = c("Clean", "Flagged"), 
    name = "Data Status"
  ) +
  facet_wrap(~ AgeGroup) +
  theme_bw() +
  labs(
    title = "QA: Force Plate Calibration & Unweighting Verification",
    subtitle = "Dashed line represents theoretical gravity (Fz = Mass * 9.81). Hover points to inspect anomalies.",
    x = "Recorded Weight (kg)",
    y = "Median Vertical Force Fz (N)"
  ) +
  theme(
    plot.title = element_text(size = 14, face = "bold"),
    plot.subtitle = element_text(size = 11, face = "italic", color = "gray30"),
    axis.title = element_text(size = 12, face = "bold"),
    legend.position = "bottom",
    strip.text = element_text(size = 11, face = "bold")
  )

# 3. Render girafe Widget
girafe(
  ggobj = qa_plot,
  options = list(
    opts_hover(css = "fill:red;stroke:black;r:6pt;"),
    opts_tooltip(css = "background-color:black;color:white;padding:10px;border-radius:5px;font-family:sans-serif;")
  )
)

Show Code

import pandas as pd
import numpy as np
import plotly.express as px

# 1. Prepare QA Data Status Formatting
QA_data = DATA.copy()
QA_data["Status"] = np.where(QA_data["flag_any"], "Flagged", "Clean")
QA_data["Flag_Symbol"] = np.where(QA_data["flag_any"], "⚠️ YES", "✅ No")

# 2. Build Interactive Scatter Plot
fig2 = px.scatter(
    QA_data,
    x="Weight",
    y="Fz_median",
    color="Status",
    facet_col="AgeGroup",
    category_orders={"AgeGroup": ["Young", "Old"], "Status": ["Clean", "Flagged"]},
    color_discrete_map={"Clean": "#1E6654", "Flagged": "#CC5500"},
    custom_data=[
        "Subject", "AgeGroup", "Gender", "Trial", 
        "Surface", "Vision", "Weight", "Fz_median", 
        "Fz_min", "duration_s", "n_samples", "Flag_Symbol"
    ],
    labels={
        "Weight": "Recorded Weight (kg)", 
        "Fz_median": "Median Vertical Force Fz (N)",
        "Status": "Data Status"
    },
    title="<b>QA: Force Plate Calibration & Unweighting Verification</b><br><sup>Dashed line represents theoretical gravity (Fz = Mass * 9.81). Hover points to inspect anomalies.</sup>"
)

# 3. Add Gravity Reference Lines (Assigned to _ so reticulate ignores it)
for col_idx in [1, 2]:
    _ = fig2.add_shape(
        type="line",
        x0=35, y0=35 * 9.81, x1=110, y1=110 * 9.81,
        line=dict(color="gray", dash="dash"),
        row=1, col=col_idx
    )

Show Code

# 4. Format Hover Tooltip & Styling (Assigned to _ so reticulate ignores it)
_ = fig2.update_traces(
    marker=dict(size=8, opacity=0.7),
    hovertemplate=(
        "<b>Subject:</b> %{customdata[0]} (%{customdata[1]}, %{customdata[2]})<br>"
        "<b>Trial:</b> %{customdata[3]}<br>"
        "<b>Condition:</b> %{customdata[4]} - %{customdata[5]}<br>"
        "<b>Recorded Weight:</b> %{customdata[6]:.1f} kg<br>"
        "<b>Median Fz:</b> %{customdata[7]:.1f} N<br>"
        "<b>Min Fz:</b> %{customdata[8]:.1f} N<br>"
        "<b>Duration:</b> %{customdata[9]}s (%{customdata[10]} samples)<br>"
        "<b>Flagged?:</b> %{customdata[11]}<extra></extra>"
    )
)

Show Code

_ = fig2.update_layout(
    template="plotly_white",
    legend=dict(orientation="h", y=-0.2, x=0.5, xanchor="center"),
    height=500
)

Show Code

# Render single figure
fig2

--- title: "Interactive Data Visualization" subtitle: "A Parallel Guide in R and Python" --- # Introduction While static graphics are ideal for publication, **interactive visualizations** are essential for data exploration and quality assurance (QA). Interactive tooltips allow researchers to inspect individual outliers, verify data flags, and examine subject metadata without cluttering the plot canvas. In this module, we demonstrate how to build interactive figures: * **In R**, we use `ggiraph`, which extends `ggplot2` syntax by converting standard geoms into interactive HTML elements. * **In Python**, we use `plotly` (`plotly.express` and `plotly.graph_objects`), which creates web-ready JavaScript widgets from `pandas` DataFrames. --- # Interactive Postural Sway Across Conditions In this first example, we make our multi-condition postural sway trajectories interactive. Hovering over any point displays subject metadata (Subject ID, Age, Gender, and Total Path Length), and hovering over an individual point highlights the subject's entire trajectory across conditions. ::: {.panel-tabset} ## R (`ggplot2` & `ggiraph`) ```{r} #| label: script03-r-plot1 #| fig-width: 9 #| fig-height: 6 library(tidyverse) library(ggiraph) # 1. Load Summary Data & Aggregate to Subject Averages DATA <- readRDS("./data/BDS/output/BDS_trial_summary.rds") %>% mutate( Subject = factor(Subject), AgeGroup = fct_relevel(AgeGroup, "Young", "Old") ) AVE_DATA <- DATA %>% select(-Trial) %>% group_by(Subject, Vision, Surface) %>% summarise( across(where(is.numeric), ~ mean(.x, na.rm = TRUE)), across(!where(is.numeric), dplyr::first), .groups = "drop" ) %>% mutate( Condition = factor(paste0(Surface, ":", Vision), levels = c("Firm:Open", "Firm:Closed", "Foam:Open", "Foam:Closed")) ) # 2. Build Tooltip Text AVE_DATA <- AVE_DATA %>% mutate(tooltip_text = paste0( "Subject ID: ", Subject, " ", "Age: ", round(Age, 1), " years ", "Gender: ", Gender, " ", "Sway (Path Length): ", round(PathLength_Total, 1), " cm" )) # 3. Define Colorblind Palette cbPalette <- c("#1E6654", "#CC5500", "#954535", "#5B7C99") # 4. Create Interactive ggplot p_sway <- ggplot(data = AVE_DATA, aes(x = Condition, y = PathLength_Total, tooltip = tooltip_text, data_id = Subject)) + geom_line(aes(group = Subject), color = "gray75", alpha = 0.5, linewidth = 0.5, position = position_dodge(width = 0.2)) + geom_point_interactive(aes(shape = Condition, col = Condition, group = Subject), size = 2.5, alpha = 0.7, position = position_dodge(width = 0.2), hover_nearest = TRUE) + facet_wrap(~ AgeGroup) + scale_color_manual(values = cbPalette) + scale_fill_manual(values = cbPalette) + theme_bw() + labs( title = "Postural Sway: Between- and Within-Subject Effects", subtitle = "Hover points to inspect subject metadata; lines connect individual trajectories", x = NULL, y = "Total Path Length (cm)", shape = "Condition", col = "Condition", fill = "Condition" ) + theme( plot.title = element_text(size = 14, face = "bold"), plot.subtitle = element_text(size = 11, face = "italic", color = "gray30"), axis.title = element_text(size = 12, face = "bold"), axis.text.x = element_blank(), axis.text.y = element_text(size = 10), legend.position = "bottom", strip.text = element_text(size = 11, face = "bold") ) # 5. Render Interactive girafe Widget girafe( ggobj = p_sway, options = list( opts_hover(css = "fill:red;stroke:black;r:5pt;"), opts_hover_inv(css = "opacity:0.2;"), opts_tooltip(css = "background-color:black;color:white;padding:8px;border-radius:4px;font-family:sans-serif;") ) ) ``` ## Python (`plotly`) ```{python} #| label: script03-py-plot1 #| jupyter_compat: true #| fig-width: 9 #| fig-height: 6 import pandas as pd import numpy as np import plotly.express as px import plotly.graph_objects as go # 1. Load Summary Data & Aggregate to Subject Averages DATA = pd.read_csv("./data/BDS/output/BDS_trial_summary.csv") DATA["AgeGroup"] = pd.Categorical(DATA["AgeGroup"], categories=["Young", "Old"], ordered=True) meta_cols = ["Subject", "Vision", "Surface", "AgeGroup", "Gender"] AVE_DATA = DATA.groupby(meta_cols, as_index=False, observed=True).mean(numeric_only=True) AVE_DATA["Condition"] = AVE_DATA["Surface"].astype(str) + ":" + AVE_DATA["Vision"].astype(str) cond_order = ["Firm:Open", "Firm:Closed", "Foam:Open", "Foam:Closed"] cbPalette = ["#1E6654", "#CC5500", "#954535", "#5B7C99"] cond_colors = {cond: cbPalette[i] for i, cond in enumerate(cond_order)} # 2. Build Base Scatter Plot fig1 = px.scatter( AVE_DATA, x="Condition", y="PathLength_Total", color="Condition", symbol="Condition", facet_col="AgeGroup", category_orders={"AgeGroup": ["Young", "Old"], "Condition": cond_order}, color_discrete_map=cond_colors, custom_data=["Subject", "Age", "Gender", "PathLength_Total"], labels={"PathLength_Total": "Total Path Length (cm)", "Condition": "Condition"}, title="Postural Sway: Between- and Within-Subject Effects Hover points to inspect subject metadata; lines connect individual trajectories" ) # 3. Add Within-Subject Trajectory Lines Across Subplots for i, age_grp in enumerate(["Young", "Old"], start=1): sub_df = AVE_DATA[AVE_DATA["AgeGroup"] == age_grp] for sub_id, grp in sub_df.groupby("Subject"): grp_sorted = grp.set_index("Condition").reindex(cond_order).dropna().reset_index() fig1.add_trace( go.Scatter( x=grp_sorted["Condition"], y=grp_sorted["PathLength_Total"], mode="lines", line=dict(color="rgba(180, 180, 180, 0.5)", width=1), showlegend=False, hoverinfo="skip" ), row=1, col=i ) # Move trajectory lines behind scatter points n_conditions = len(cond_order) fig1.data = tuple(list(fig1.data[n_conditions:]) + list(fig1.data[:n_conditions])) # 4. Format Custom Hover Tooltip & Layout fig1.update_traces( selector=dict(mode="markers"), marker=dict(size=8, opacity=0.8), hovertemplate=( "Subject ID: %{customdata[0]} " "Age: %{customdata[1]:.1f} years " "Gender: %{customdata[2]} " "Sway (Path Length): %{customdata[3]:.1f} cm<extra></extra>" ) ) fig1.update_layout( template="plotly_white", legend=dict(orientation="h", y=-0.2, x=0.5, xanchor="center"), height=500 ) # Single final expression evaluated fig1 ``` ::: --- # Interactive Quality Assurance (Force Plate Calibration & Unweighting) In this second example, we evaluate force plate calibration and screen for potential artifacts (such as a participant stepping off or unweighting the platform). Under static gravity, vertical force $F_z$ should approximate body weight in Newtons: $$F_z \approx \text{Mass (kg)} \times 9.81 \text{ m/s}^2$$ Hovering over flagged data points instantly reveals trial parameters, recorded body weight, and minimum vs. median $F_z$ values. ::: {.panel-tabset} ## R (`ggplot2` & `ggiraph`) ```{r} #| label: script03-r-plot2 #| fig-width: 9 #| fig-height: 6 # 1. Prepare Interactive Tooltips with Detailed QA Information QA_data <- DATA %>% mutate( QA_tooltip = paste0( "Subject: ", Subject, " (", AgeGroup, ", ", Gender, ") ", "Trial: ", Trial, " ", "Condition: ", Surface, " - ", Vision, " ", "Recorded Weight: ", Weight, " kg ", "Median Fz: ", round(Fz_median, 1), " N ", "Min Fz: ", round(Fz_min, 1), " N ", "Duration: ", duration_s, "s (", n_samples, " samples) ", "Flagged?: ", ifelse(flag_any, "⚠️ YES", "✅ No") ) ) # 2. Build Interactive Scatter Plot qa_plot <- ggplot(QA_data, aes(x = Weight, y = Fz_median, tooltip = QA_tooltip, data_id = Trial)) + geom_abline(slope = 9.81, intercept = 0, linetype = "dashed", color = "gray50") + geom_point_interactive(aes(color = flag_any), size = 2.5, alpha = 0.7) + scale_color_manual( values = c("FALSE" = "#1E6654", "TRUE" = "#CC5500"), labels = c("Clean", "Flagged"), name = "Data Status" ) + facet_wrap(~ AgeGroup) + theme_bw() + labs( title = "QA: Force Plate Calibration & Unweighting Verification", subtitle = "Dashed line represents theoretical gravity (Fz = Mass * 9.81). Hover points to inspect anomalies.", x = "Recorded Weight (kg)", y = "Median Vertical Force Fz (N)" ) + theme( plot.title = element_text(size = 14, face = "bold"), plot.subtitle = element_text(size = 11, face = "italic", color = "gray30"), axis.title = element_text(size = 12, face = "bold"), legend.position = "bottom", strip.text = element_text(size = 11, face = "bold") ) # 3. Render girafe Widget girafe( ggobj = qa_plot, options = list( opts_hover(css = "fill:red;stroke:black;r:6pt;"), opts_tooltip(css = "background-color:black;color:white;padding:10px;border-radius:5px;font-family:sans-serif;") ) ) ``` ## Python (`plotly`) ```{python} #| label: script03-py-plot2 #| jupyter_compat: true #| fig-width: 9 #| fig-height: 6 import pandas as pd import numpy as np import plotly.express as px # 1. Prepare QA Data Status Formatting QA_data = DATA.copy() QA_data["Status"] = np.where(QA_data["flag_any"], "Flagged", "Clean") QA_data["Flag_Symbol"] = np.where(QA_data["flag_any"], "⚠️ YES", "✅ No") # 2. Build Interactive Scatter Plot fig2 = px.scatter( QA_data, x="Weight", y="Fz_median", color="Status", facet_col="AgeGroup", category_orders={"AgeGroup": ["Young", "Old"], "Status": ["Clean", "Flagged"]}, color_discrete_map={"Clean": "#1E6654", "Flagged": "#CC5500"}, custom_data=[ "Subject", "AgeGroup", "Gender", "Trial", "Surface", "Vision", "Weight", "Fz_median", "Fz_min", "duration_s", "n_samples", "Flag_Symbol" ], labels={ "Weight": "Recorded Weight (kg)", "Fz_median": "Median Vertical Force Fz (N)", "Status": "Data Status" }, title="QA: Force Plate Calibration & Unweighting Verification Dashed line represents theoretical gravity (Fz = Mass * 9.81). Hover points to inspect anomalies." ) # 3. Add Gravity Reference Lines (Assigned to _ so reticulate ignores it) for col_idx in [1, 2]: _ = fig2.add_shape( type="line", x0=35, y0=35 * 9.81, x1=110, y1=110 * 9.81, line=dict(color="gray", dash="dash"), row=1, col=col_idx ) # 4. Format Hover Tooltip & Styling (Assigned to _ so reticulate ignores it) _ = fig2.update_traces( marker=dict(size=8, opacity=0.7), hovertemplate=( "Subject: %{customdata[0]} (%{customdata[1]}, %{customdata[2]}) " "Trial: %{customdata[3]} " "Condition: %{customdata[4]} - %{customdata[5]} " "Recorded Weight: %{customdata[6]:.1f} kg " "Median Fz: %{customdata[7]:.1f} N " "Min Fz: %{customdata[8]:.1f} N " "Duration: %{customdata[9]}s (%{customdata[10]} samples) " "Flagged?: %{customdata[11]}<extra></extra>" ) ) _ = fig2.update_layout( template="plotly_white", legend=dict(orientation="h", y=-0.2, x=0.5, xanchor="center"), height=500 ) # Render single figure fig2 ``` :::