Improving Visualization of Multiple Data Points in a Scatter Chart

You’re hitting two Power BI scatter limits:

1. Bubble size is scaled non-linearly and then capped (so close values often look the same).

2. Your current formula EXP([pax]/Max * 2) actually compresses most points unless the ratio is near 1.

1) Use a stronger (and controllable) size scaling

A good pattern is normalize → power curve → scale to a range:

Dot Size =
VAR v = [0PaxCount]
VAR vMin = CALCULATE( MIN([0PaxCount]), ALLSELECTED(MVPI) )
VAR vMax = CALCULATE( MAX([0PaxCount]), ALLSELECTED(MVPI) )
VAR n = DIVIDE( v - vMin, vMax - vMin )          -- 0..1
VAR curved = POWER( n, 0.35 )                    -- <1 expands differences
RETURN
1 + 99 * curved                                  -- force range 1..100

2) Color by size (discrete “bands”) using conditional formatting

Power BI scatter can’t do a continuous gradient color by a measure directly like some visuals, but you can do it with binning:

Create a band measure/column:

Size Band =
VAR n =
    DIVIDE(
        [0PaxCount],
        CALCULATE( MAX([0PaxCount]), ALLSELECTED(MVPI) )
    )
RETURN
SWITCH(
    TRUE(),
    n < 0.2, "1",
    n < 0.4, "2",
    n < 0.6, "3",
    n < 0.8, "4",
    "5"
)

Then put Size Band on Legend and assign colors (light → dark).

3) If outliers exist: cap the max so mid values separate better

Outliers make all other bubbles look similar.

If you can add a helper measure, cap vMax using a percentile (approximation approach):

• Create a “Max excluding top x%” value in Power Query is best, but if you must stay DAX-only, easiest is: use a fixed cap (business-defined), or use a measure like Median/Avg based cap.

Example with a business cap:

VAR vMax = MIN(
    CALCULATE( MAX([0PaxCount]), ALLSELECTED(MVPI) ),
    5000  -- cap
)

Hi YYlee 

In Power BI’s native scatter chart there is no discrete control over bubble size per category. Bubble size is always treated as a continuous, normalized scale, with internal min/max limits applied by the visual itself. As a result, even when using nonlinear DAX transformations (EXP, LOG, POWER, etc.), small-to-medium differences tend to be visually compressed and remain hard to distinguish.

This is therefore not a DAX limitation but a visualization constraint of the scatter chart.

More importantly, from a data visualization perspective, scatter charts with varying bubble sizes are generally not suitable for comparing magnitudes. Human perception is poor at comparing areas, especially circular areas, and even more so when values are close (e.g. 3 vs. 4 or 5 vs. 6). This makes precise or even relative comparison unreliable, regardless of how the size is calculated.

If the goal is to compare similarity or differences between categories, a table combined with bar visuals (e.g. data bars, column or bar charts) will usually be far more effective and interpretable.

If the goal is instead to analyze spatial patterns, clustering, or directional behavior, a scatter chart can still be appropriate, but bubble size should be used very cautiously (or reduced to a very coarse scale), and color or faceting may communicate the message more clearly.

Clarifying the analytical goal—what the viewer is expected to understand or decide based on the chart—would make it easier to recommend a more effective visual approach.

If this post helps, then please consider Accepting it as the solution to help the other members find it more quickly

Hii YYlee 

Instead of exponential growth, we will usePower Scaling to create a steeper visual gap and Conditional Formatting to add a second layer of identification.

​Step 1: Update the Size Measure

​Replace your EXP logic with a POWER function. Cubing the value (x^3) creates a much more dramatic size difference than EXP for normalized values.

0Dot Size Adjusted = 
VAR MaxPax = CALCULATE ( MAX ( [0PaxCount] ), ALLSELECTED ( MVPI ) )
VAR NormalizedValue = DIVIDE ( [0PaxCount], MaxPax )
RETURN
// Raising to the power of 3 creates a high-contrast visual gap
POWER(NormalizedValue, 3) * 100

Step 2: Apply a Color Gradient (Heatmap Effect)

​Since eyes detect color changes faster than slight size changes, apply a gradient based on the count.

1. ​Go to the Format pane of your Scatter Chart.
2. ​Navigate to Markers > Color.
3. ​Click the fx button (Conditional Formatting).
4. ​Set Format style to Gradient.
5. ​Base it on the field [0PaxCount].
6. ​Set the Minimum to a light color (e.g., Light Blue) and Maximum to a dark color (e.g., Dark Navy).

​Step 3: Enable Transparency

​To help with overlapping dots (3 & 4 / 5 & 6), set the Marker Transparency to 25%.

• ​Why? When two dots overlap, the intersection will appear darker, immediately signaling that multiple data points are clustered there.

​Why this works:

• ​Mathematical Contrast: Power scaling forces the larger values to be significantly larger than the median values, making them easier to tell apart.
• ​Dual-Encoding: By using both Size and Color to represent the same value, you complement the visual data, making the report more accessible for everyone.

​Summary for the Community

​If your scatter chart bubbles look too similar, try moving to a POWER function and always use a Color Gradient to make the "clusters" stand out.

​If these adjustments make your dots easier to distinguish, please mark this as the "Accepted Solution"!