Axis titles, labels, and hover text

The idea

A default chart speaks in column names: revenue, order_date, qty_sold. Your audience speaks in "Revenue (Rp)", "Order date" and "Units sold". The difference between a chart people glance at and one they trust is often just labels, number formats and a good hover box. All three take a line or two.

import pandas as pd
import plotly.express as px

sales = pd.read_csv("sales.csv")
completed = sales[sales["status"] == "completed"]
monthly = (completed.assign(month=completed["order_date"].str[:7])
           .groupby("month", as_index=False)
           .agg(revenue=("revenue", "sum"), orders=("order_id", "nunique")))
print(monthly)

     month  revenue  orders
0  2024-01  1693000       6
1  2024-02  1752000       4
2  2024-03  1027000       2
3  2024-04  3987000       7
4  2024-05  2552000       5
5  2024-06   825000       2

labels=: rename columns for the chart

Express uses column names as axis titles, legend titles and hover labels. labels= maps column names to display names, everywhere at once:

fig = px.bar(monthly, x="month", y="revenue",
             labels={"month": "Month", "revenue": "Revenue (Rp)"},
             title="Completed revenue per month")
fig.show()
print(fig.layout.xaxis.title.text, "|", fig.layout.yaxis.title.text)

Month | Revenue (Rp)

Prefer labels= over renaming DataFrame columns just for a chart: your data keeps code-friendly names, and the chart gets human ones.

You can also set titles after creating the figure, which works for Graph Objects figures too:

fig.update_xaxes(title_text="Month (2024)")
fig.update_yaxes(title_text="Revenue (Rupiah)")

Number formats on the axis

Plotly writes 4,000,000 as 4M by default. For a Rupiah audience, you might want separators and a prefix. tickformat uses the same mini-language as the text_auto formats in the previous module:

fig = px.bar(monthly, x="month", y="revenue",
             labels={"month": "Month", "revenue": "Revenue"})
fig.update_yaxes(tickprefix="Rp ", tickformat=",")
fig.show()
print(repr(fig.layout.yaxis.tickprefix), repr(fig.layout.yaxis.tickformat))

'Rp ' ','

tickprefix and ticksuffix add text before or after each tick label, like "Rp " or " units".

Category order

Axis categories appear in the order of the data, or alphabetically. Force an order with category_orders:

by_channel = completed.groupby("channel", as_index=False)["revenue"].sum()
fig = px.bar(by_channel, x="channel", y="revenue",
             category_orders={"channel": ["web", "app"]})
fig.show()

Useful for natural orders that aren't alphabetical: weekdays, sizes (S, M, L), ratings, or "Q1, Q2, Q3, Q4".

Hover text: what appears on mouse-over

The hover box is where interactive charts beat static ones. Express shows the x and y values by default. Add more context with hover_name (bold title) and hover_data (extra rows):

fig = px.scatter(monthly, x="orders", y="revenue", hover_name="month",
                 hover_data={"orders": True, "revenue": ":,"},
                 labels={"orders": "Orders", "revenue": "Revenue (Rp)"})
fig.show()

• hover_name="month" puts the month in bold at the top.
• hover_data is a dict: True shows a column, False hides one, and a format string like ":," formats it (the colon starts the format).

hovertemplate: full control

For complete control, write the hover text yourself with hovertemplate. Values are inserted with %{...}:

fig = px.bar(monthly, x="month", y="revenue")
fig.update_traces(hovertemplate="<b>%{x}</b><br>Revenue: Rp %{y:,}<extra></extra>")
fig.show()
print(fig.data[0].hovertemplate)

<b>%{x}</b><br>Revenue: Rp %{y:,}<extra></extra>

• %{x} and %{y:,} insert values, with an optional format after the colon.
• <br> is a line break, <b>...</b> bold.
• <extra></extra> removes the grey trace-name box that otherwise appears beside the hover.

To show a column that isn't on an axis, pass it as custom_data in Express and reference it as %{customdata[0]}:

fig = px.bar(monthly, x="month", y="revenue", custom_data=["orders"])
fig.update_traces(hovertemplate="%{x}: Rp %{y:,} from %{customdata[0]} orders<extra></extra>")
fig.show()

Why this matters for clean code

Labels are the user interface of your analysis. A chart titled "revenue by month" with an axis reading 4M invites "4 million what?" A clean chart answers the questions before they're asked: what is measured, in which unit, over what period. Keep the formatting in a few explicit lines after the px call, so the "what" (data) and the "how it looks" (labels) stay separate and easy to change.

Recap

• labels={"col": "Nice name"} renames columns in axes, legends and hover at once.
• update_xaxes(title_text=...) / update_yaxes(...) set titles on any figure.
• tickformat (",", ".2s", ".1%"), tickprefix, ticksuffix format axis numbers; separators=",." switches to Indonesian style.
• category_orders={"col": [...]} fixes category order.
• hover_name, hover_data add context; hovertemplate with %{x}, %{y:,}, %{customdata[0]} gives full control.