Skip to main content

Scatterplots and Histograms

VisiData renders scatterplots by plotting one numeric column against another, and histograms through the frequency table's built-in histogram column. Both run entirely in the terminal with no external libraries.

Learning Focus

Scatterplots reveal correlation between two variables. Histograms reveal distribution shape. Learn when each is more informative than a simple frequency table.

Sample Data Used in This Lesson​

mkdir -p ~/github/practice-folder/visidata/09-metrics

cat > ~/github/practice-folder/visidata/09-metrics/requests.csv << 'EOF'
method,status,response_ms,bytes_sent
GET,200,45,1234
GET,200,120,8900
POST,201,320,456
GET,404,12,98
GET,200,980,45678
POST,500,4500,102
GET,200,67,2300
DELETE,204,38,0
GET,200,1800,12000
POST,500,3200,89
EOF

vd ~/github/practice-folder/visidata/09-metrics/requests.csv

Scatterplot: Two Numeric Variables​

Goal: Visualize if larger byte sizes cause longer response times.

Step 1 — Cast and key the x-axis:

# Move to 'response_ms' column
#        # cast to integer
!        # mark as key column (x-axis)

Step 2 — Cast the y-axis column:

# Move to 'bytes_sent' column
#        # cast to integer

Step 3 — Plot:

.        # dot → opens canvas sheet

Canvas result (ASCII approximation):

bytes_sent vs response_ms
│
50000│              ·
40000│
30000│
20000│
10000│          ·
     │ ·· ·                   ·
     │                        ·    ·
     └──────────────────────────────
       0    500   1000  2000  5000
             response_ms

Trend visible: high response_ms correlates with a 500 error (outliers top-right area).

Color-Coded by Category​

When a categorical key column is set alongside a numeric key, VisiData assigns distinct colors per category:

# Move to 'method' column
!        # categorical key (distinct colors per method)

# Move to 'response_ms' column
!        # numeric key (x-axis)

# Move to 'bytes_sent' column
.        # scatterplot: x=response_ms, each method a different color

Each HTTP method (GET, POST, DELETE) appears in a distinct color on the canvas — pattern: POST requests have longer response times (higher response_ms values).

Histogram from Frequency Table​

The frequency table automatically includes a text-based histogram column using ■ characters:

# Move to 'status' column
Shift+F

Result — text histogram built-in:

status  count  percent   histogram
200     6      60.0%     ████████████████████
201     1      10.0%     ███
204     1      10.0%     ███
404     1      10.0%     ███
500     2      20.0%     ██████

This is already a histogram — readable without a canvas.

Canvas Histogram from Frequency Table​

For a canvas-based (graphical) histogram:

# In the frequency table (after Shift+F on 'status'):
# Move cursor to 'count' column
.        # plot count per status as a canvas histogram

Canvas shows count bars per status code visually.

Response Time Distribution​

vd ~/github/practice-folder/visidata/09-metrics/requests.csv

# Cast response_ms to int
#        # on 'response_ms' column

# Open frequency table to see distribution
Shift+F

Frequency table shows how response times distribute — are most requests fast (< 100ms) or slow (> 1000ms)?

Canvas Navigation Keys​

KeyAction
+ / -Zoom in / out
_Zoom to fit all data
h j k lPan the canvas
sSelect source rows under canvas cursor
EnterOpen source rows for canvas cursor area
vToggle graph labels
1–9Toggle individual plot layers (for multi-column overlays)
xSet x-axis range manually
ySet y-axis range manually

Reading the Canvas​

·  or  ⠁⠂⠄⠈   sparse data points (few in this cell)
⠿  or  ⣿       dense cluster of many points
color           distinct categorical values (when categorical key set)
─   │           axis lines
axis labels     scale ticks

Troubleshooting​

ProblemCauseFix
Scatterplot shows vertical linesX-axis column is categoricalUse a numeric column as the key
Canvas all one colorNo categorical key setMark a categorical column as an additional key
Points too sparse to seeData range too wideUse x and y to set axis ranges
Canvas renders slowlyVery many distinct pointsFilter to a sample first
■ histogram too small to readTerminal too narrowWiden terminal or use canvas graph

Hands-On Practice​

vd ~/github/practice-folder/visidata/09-metrics/requests.csv

# 1. Cast response_ms to int: #
# 2. Mark response_ms as key: !
# 3. Cast bytes_sent to int: #
# 4. Press . → scatterplot of bytes_sent vs response_ms
# 5. Press + → zoom in to see the outlier points
# 6. Press Enter → see source rows for that area
# 7. Press q → back to canvas
# 8. Press q → back to source sheet

# 9. Move to 'status' column → Shift+F → view text histogram
# 10. Move cursor to 'count' column → press . → canvas histogram

What's Next​