import marimo
__generated_with = "0.21.1"
app = marimo.App(width="medium")
@app.cell
async def _():
import sys
import io
import math
import marimo as mo
import pandas as pd
import openpyxl
return io, math, mo, pd
@app.cell
def _(mo):
# Buttons to upload the data files required for the visualisation
upload_yield = mo.ui.file(label="Upload: Coffee_yield.xlsx", kind="button", multiple=False)
upload_species = mo.ui.file(label="Upload: Plant_species_and_average...xlsx", kind="button", multiple=False)
upload_decomp = mo.ui.file(label="Upload: Total_species_composition.xlsx", kind="button", multiple=False)
# Assign to variable and display the UI
upload_ui = mo.vstack([
mo.md("### Please provide the correct data files to view the visual"),
upload_yield,
upload_species,
upload_decomp
], align="center")
# Place as the last statement to ensure Marimo renders it!
upload_ui
return upload_decomp, upload_species, upload_yield, upload_ui
@app.cell
def _(io, mo, pd, upload_decomp, upload_species, upload_yield):
# Exectution of this cell and everything below is paused until all files are uploaded
mo.stop(
not upload_yield.value or not upload_species.value or not upload_decomp.value,
mo.md("*Waiting for all three files to be uploaded...*")
)
###### PREPARATION ######
# Read the uploaded files from browser memory
df_yield = pd.read_excel(io.BytesIO(upload_yield.value[0].contents))
df_species = pd.read_excel(io.BytesIO(upload_species.value[0].contents))
df_decomposition = pd.read_excel(io.BytesIO(upload_decomp.value[0].contents))
# Hardcoded column names used in the data files
COL_SPECIES_NAME = "Species name"
COL_SPECIES_GROUP = "Species group"
COL_DECOMP_SPECIES = df_decomposition.columns[0]
# Build a site -> yield lookup from Coffee_yield.xlsx
site_yield_map = dict(zip(df_yield["Site ID"].astype(str), df_yield["Mean_CC_Yield"]))
# In `Plant_species_and_average_yield.xlsx`, empty group cells are filled
df_species[COL_SPECIES_GROUP] = df_species[COL_SPECIES_GROUP].ffill()
GROUPS = df_species[COL_SPECIES_GROUP].dropna().unique().tolist()
# In `Total_species_decomposition.xlsx`, set index to species for easy row lookups
df_decomposition.set_index(COL_DECOMP_SPECIES, inplace=True)
ALL_SITES = [str(col) for col in df_decomposition.columns]
# Build the species_data dictionary >> dictionary used to build the visual later
species_data = []
for idx, row in df_species.iterrows():
sp_id = str(row[COL_SPECIES_NAME])
group = str(row[COL_SPECIES_GROUP]) if COL_SPECIES_GROUP in df_species.columns else GROUPS[idx % len(GROUPS)]
# First: find which sites this species occurs in
present_in = []
if sp_id in df_decomposition.index:
species_row = df_decomposition.loc[sp_id]
sites_with_species = species_row[species_row == 1]
present_in = sites_with_species.index.astype(str).tolist()
# Then: compute mean yield across those sites using Coffee_yield.xlsx
site_yields = [site_yield_map[site] for site in present_in if site in site_yield_map]
avg_yield = sum(site_yields) / len(site_yields) if site_yields else 0.0
species_data.append({
"id": sp_id,
"group": group,
"yield": avg_yield,
"num_sites": len(present_in),
"sites": present_in
})
# Sort species from most common (center) to least common (edge)
species_data.sort(key=lambda x: x["num_sites"], reverse=True)
return ALL_SITES, GROUPS, species_data
@app.cell
def _(ALL_SITES, mo, species_data):
# Create UI controls
total_species = len(species_data)
slider_count = mo.ui.slider(
start=5, stop=total_species, step=1, value=int(total_species/2), label="Species shown:"
)
drop_single = mo.ui.dropdown(
options=ALL_SITES, value=ALL_SITES[0] if ALL_SITES else "", label="Highlight Site:"
)
drop_comp1 = mo.ui.dropdown(
options=ALL_SITES, value=ALL_SITES[0] if ALL_SITES else "", label="Compare Site 1:"
)
drop_comp2 = mo.ui.dropdown(
options=ALL_SITES, value=ALL_SITES[1] if len(ALL_SITES) > 1 else ALL_SITES[0], label="Compare Site 2:"
)
return drop_comp1, drop_comp2, drop_single, slider_count
@app.cell
def _(mo):
dropdown_styles = mo.Html("""
""")
tabs = mo.ui.tabs({
"General Overview": mo.md("*Viewing all species colored by their primary group.*"),
"Individual Site": mo.md("*Select a site using the dropdown below.*"),
"Compare Sites": mo.md("*Select two sites to compare using the dropdowns below.*"),
})
return dropdown_styles, tabs
@app.cell
def _(
drop_comp1,
drop_comp2,
drop_single,
dropdown_styles,
mo,
slider_count,
tabs,
):
_active = tabs.value
if _active == "Individual Site":
_site_selector = mo.hstack([drop_single], justify="center")
elif _active == "Compare Sites":
_site_selector = mo.hstack([drop_comp1, drop_comp2], gap=4)
else:
_site_selector = mo.Html("")
controls = mo.vstack([
dropdown_styles,
tabs,
_site_selector,
mo.hstack([slider_count], justify="center")
], align="center", gap=4)
return (controls,)
@app.cell
def _(math, tabs):
# Sunburst chart: dimensions + position
CX, CY = 500, 380
MAX_RADIUS = 350
MIN_RADIUS = 80
TW, TH = 245, 105
# Colors
preferred_hues = [30, 120, 210]
SITE_LEGEND_BKG = "#1e1e2e"
SITE_LEGEND_BORDER = "#45475a"
SITE_LEGEND_TEXT = "#cdd6f4"
SITE_COLOR_S1_ONLY = "#f5b0c6"
SITE_COLOR_S2_ONLY = "#d8b4fe"
SITE_COLOR_BOTH = "#f9e2af"
SITE_COLOR_NEITHER = "#e0e0e0"
# Helper functions
def polar_to_cartesian(cx, cy, r, angle_deg):
rad = math.radians(angle_deg)
return cx + r * math.cos(rad), cy + r * math.sin(rad)
def build_arc(cx, cy, r_inner, r_outer, start_angle, end_angle):
if end_angle - start_angle <= 0.05:
end_angle = start_angle + 0.05
p1 = polar_to_cartesian(cx, cy, r_outer, start_angle)
p2 = polar_to_cartesian(cx, cy, r_outer, end_angle)
p3 = polar_to_cartesian(cx, cy, r_inner, end_angle)
p4 = polar_to_cartesian(cx, cy, r_inner, start_angle)
large_arc = "0" if end_angle - start_angle <= 180 else "1"
return f"M {p1[0]} {p1[1]} A {r_outer} {r_outer} 0 {large_arc} 1 {p2[0]} {p2[1]} L {p3[0]} {p3[1]} A {r_inner} {r_inner} 0 {large_arc} 0 {p4[0]} {p4[1]} Z"
def make_tooltip(unique_id, tx, ty, group_name, species_name, avg_yield, num_sites):
# Bound the tooltip coordinates so it doesn't clip off the 1000x1000 SVG canvas
tx = max(10, min(tx, 1000 - TW - 10))
ty = max(10, min(ty, 1000 - TH - 10))
safe_group = group_name.strip().upper()[:40]
safe_name = species_name.strip()[:38]
yield_str = f"Average yield: {avg_yield:.1f} kg ha\u207b\u00b9"
sites_str = f"Occurs in: {num_sites} site(s)"
tip_id = unique_id.replace("seg", "tip")
return f"""
{safe_group}{safe_name}{yield_str}{sites_str}"""
_ = tabs
return (
CX,
CY,
MAX_RADIUS,
MIN_RADIUS,
SITE_COLOR_BOTH,
SITE_COLOR_NEITHER,
SITE_COLOR_S1_ONLY,
SITE_COLOR_S2_ONLY,
SITE_LEGEND_BKG,
SITE_LEGEND_BORDER,
SITE_LEGEND_TEXT,
TH,
TW,
build_arc,
make_tooltip,
polar_to_cartesian,
preferred_hues,
)
@app.cell
def _(GROUPS, preferred_hues, slider_count, species_data, tabs):
# Filtering data + active tab
active_data = species_data[:slider_count.value]
active_tab = tabs.value
group_hues = {
group_name: preferred_hues[i % len(preferred_hues)]
for i, group_name in enumerate(GROUPS)
}
grouped_data = {g: [] for g in GROUPS}
for s in active_data:
grouped_data[s["group"]].append(s)
return active_tab, group_hues, grouped_data
@app.cell
def _(
CX,
CY,
GROUPS,
MAX_RADIUS,
MIN_RADIUS,
SITE_COLOR_BOTH,
SITE_COLOR_NEITHER,
SITE_COLOR_S1_ONLY,
SITE_COLOR_S2_ONLY,
TH,
TW,
active_tab,
build_arc,
drop_comp1,
drop_comp2,
drop_single,
group_hues,
grouped_data,
make_tooltip,
polar_to_cartesian,
):
# Build the core sunburst paths
core_paths = []
tooltip_elements = []
css_hover_rules = []
tab_id_prefix = active_tab.replace(" ", "_").lower() # unique tab prefix
seg_index = 0 # unique id counter shared across all segments
for group_idx, group_name in enumerate(GROUPS):
group_items = grouped_data[group_name]
if not group_items:
continue
base_angle = group_idx * 120 # hardcoded 120 degree angle bcs only 3 groups
tiers = {}
for item in group_items:
sites = item["num_sites"]
if sites not in tiers:
tiers[sites] = []
tiers[sites].append(item)
sorted_tier_keys = sorted(tiers.keys(), reverse=True)
num_tiers = len(sorted_tier_keys)
ring_thickness = (MAX_RADIUS - MIN_RADIUS) / max(1, num_tiers)
for tier_idx, sites_key in enumerate(sorted_tier_keys):
tier_items = tiers[sites_key]
r_inner = MIN_RADIUS + (tier_idx * ring_thickness) # calculate segment properties (width / radius / ...)
r_outer = r_inner + ring_thickness
r_center = (r_inner + r_outer) / 2
total_yield = sum(item["yield"] for item in tier_items)
current_start_angle = base_angle
for item in tier_items:
unique_seg_id = f"{tab_id_prefix}-seg-{seg_index}"
unique_tip_id = f"{tab_id_prefix}-tip-{seg_index}"
if total_yield > 0:
sweep_angle = (item["yield"] / total_yield) * 120
else:
sweep_angle = 120 / len(tier_items)
end_angle = current_start_angle + sweep_angle
opacity = 1.0
fill_color = ""
if active_tab == "Compare Sites": # coloring depends on which tab the user is viewing
in_s1 = drop_comp1.value in item["sites"]
in_s2 = drop_comp2.value in item["sites"]
if in_s1 and in_s2: fill_color = SITE_COLOR_BOTH
elif in_s1: fill_color = SITE_COLOR_S1_ONLY
elif in_s2: fill_color = SITE_COLOR_S2_ONLY
else: fill_color = SITE_COLOR_NEITHER
elif active_tab == "Individual Site":
hue = group_hues[group_name]
lightness = 85 - min(50, (item["yield"] / max(1, total_yield)) * 45)
fill_color = f"hsl({hue}, 70%, {lightness}%)"
if drop_single.value not in item["sites"]:
opacity = 0.15
else:
hue = group_hues[group_name]
lightness = 85 - min(50, (item["yield"] / max(1, total_yield)) * 45)
fill_color = f"hsl({hue}, 70%, {lightness}%)"
path_d = build_arc(CX, CY, r_inner, r_outer, current_start_angle, end_angle)
angle_center = current_start_angle + sweep_angle / 2
core_paths.append(
f''
)
tip_r = r_outer + 20
tip_cx, tip_cy = polar_to_cartesian(CX, CY, tip_r, angle_center)
tx = tip_cx if tip_cx < CX else tip_cx - TW
ty = tip_cy - TH / 2
tooltip_elements.append(
make_tooltip(unique_seg_id, tx, ty, group_name,
item["id"], item["yield"], item["num_sites"])
)
# add hovering effect
css_hover_rules.append(
f"svg:has(#seg-{unique_seg_id}:hover) #seg-{unique_seg_id} "
f"{{ filter: brightness(1.35) drop-shadow(0 0 7px rgba(255,255,255,0.6)); }}\n"
f"svg:has(#seg-{unique_seg_id}:hover) #{unique_tip_id} "
f"{{ visibility: visible; }}"
)
current_start_angle = end_angle
seg_index += 1
return core_paths, css_hover_rules, tooltip_elements
@app.cell
def _(
CX,
CY,
MAX_RADIUS,
MIN_RADIUS,
SITE_COLOR_BOTH,
SITE_COLOR_NEITHER,
SITE_COLOR_S1_ONLY,
SITE_COLOR_S2_ONLY,
SITE_LEGEND_BKG,
SITE_LEGEND_BORDER,
SITE_LEGEND_TEXT,
active_tab,
drop_comp1,
drop_comp2,
polar_to_cartesian,
):
# Build annotations and legends
annotation_elements = []
if active_tab == "Compare Sites": # legend for Compare Sites tab
site1_label = drop_comp1.value
site2_label = drop_comp2.value
legend_items = [
(SITE_COLOR_S1_ONLY, f"Only in {site1_label}"),
(SITE_COLOR_S2_ONLY, f"Only in {site2_label}"),
(SITE_COLOR_BOTH, f"In both sites"),
(SITE_COLOR_NEITHER, "In neither site"),
]
LX, LY = 0, 0 # top-left corner of the legend box
LW, LH = 210, 150 # box dimensions
ROW_H = 26 # vertical spacing between rows
SWATCH_S = 14 # swatch square size
annotation_elements.append(
f''
f''
f''
f'SITE COMPARISON'
f''
)
for i, (color, label) in enumerate(legend_items):
row_y = LY + 42 + i * ROW_H
annotation_elements.append(
f''
f'{label}'
)
annotation_elements.append('')
# dividers
for i in range(3):
angle = i * 120
p1 = polar_to_cartesian(CX, CY, MIN_RADIUS, angle)
p2 = polar_to_cartesian(CX, CY, MAX_RADIUS, angle)
annotation_elements.append(
f''
)
# group labels
lbl_1 = polar_to_cartesian(CX, CY, MAX_RADIUS + 25, 60)
lbl_2 = polar_to_cartesian(CX, CY, MAX_RADIUS + 25, 180)
lbl_3 = polar_to_cartesian(CX, CY, MAX_RADIUS + 25, 300)
annotation_elements.extend([
f'Woody vascular plants',
f'Non-woody vascular plants',
f'Bryophytes',
])
return (annotation_elements,)
@app.cell
def _(active_tab, drop_single, mo, species_data):
# Build the recommendations panel for the Individual Site tab
recommendations_panel = mo.Html("")
if active_tab == "Individual Site" and drop_single.value:
selected_site = drop_single.value
present = [s for s in species_data if selected_site in s["sites"]]
not_present = [s for s in species_data if selected_site not in s["sites"]]
# Top 5 to REMOVE: species present at site with the lowest avg yield
to_remove = sorted(present, key=lambda x: x["yield"])[:5]
# Top 5 to ADD: species not present at site with the highest avg yield
to_add = sorted(not_present, key=lambda x: x["yield"], reverse=True)[:5]
def make_row(rank, species, color):
return f"""
#{rank}
{species['id'][:35]}
{species['group']} · occurs in {species['num_sites']} site(s)
{species['yield']:.1f} kg ha⁻¹
"""
remove_rows = "".join(make_row(i+1, s, "#f38ba8") for i, s in enumerate(to_remove))
add_rows = "".join(make_row(i+1, s, "#a6e3a1") for i, s in enumerate(to_add))
recommendations_panel = mo.Html(f"""