During an internal model comparison, you need to show leadership that Model B has a slightly higher F1-score than Model A (0.904 versus 0.889). A junior analyst suggests a bar chart whose y-axis starts at 0.88 so the difference looks dramatic. To comply with visualization best practices and avoid unintentionally deceptive reporting, which revision should you recommend?
Replace the bars with a dual-axis combo chart, plotting Model A on the left y-axis and Model B on the right y-axis so each series fills the graph vertically.
Keep the current axis limits but convert the bars to a 3-D column chart so the height difference is even easier to see.
Redraw the bar chart with a y-axis that starts at zero and place the exact F1-score values above each bar.
Apply a logarithmic scale to the y-axis and round each score to two decimal places to emphasize the percentage change.
Bar charts encode value by the length of the bar, so viewers assume the bar's base represents zero. Truncating the y-axis inflates perceived differences and can mislead decision-makers. The safest fix is to redraw the bars so the y-axis begins at zero-maintaining an honest scale-and, if desired, add data labels so small numeric differences are still obvious. Switching to 3-D effects, dual independent axes, or a logarithmic scale would each introduce additional visual distortions without solving the underlying problem.
Ask Bash
Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.
Why should a bar chart's y-axis start at zero?
Open an interactive chat with Bash
What is an F1-score, and why is it important in model comparison?
Open an interactive chat with Bash
How is a logarithmic scale different from a linear scale in data visualization?