Dual-Axis Charts: Plot Two Scales with twinx()

Some comparisons span two scales at once. Runs scored count in the hundreds; win percentage is a fraction hovering near 0.5. Plot them on the same y-axis and the win-percentage line flatlines along the bottom, invisible. A second y-axis via ax.twinx() fixes that — one chart, two scales, both readable. It's a genuinely useful tool, and also one of the easiest ways to mislead, so we'll cover both.

This builds on Your First Visualization. The data is the bundled sample_standings.csv (real 2023 MLB standings), so it runs offline.

1. Two columns, two scales

   We'll compare each top team's runs scored (hundreds) against its win percentage (a fraction). Their ranges barely overlap — exactly the situation a dual axis is built for.

   import pandas as pd
   
   df = pd.read_csv("sample_standings.csv")
   top = df.sort_values("W", ascending=False).head(10).reset_index(drop=True)
   print(top[["Abbr", "RS", "WinPct"]].to_string())

   Runs scored vs win percentage

     Abbr   RS  WinPct
   0  ATL  947   0.642
   1  BAL  807   0.623
   2  LAD  906   0.617
   3   TB  860   0.611
   4  MIL  728   0.568
   5  HOU  827   0.556
   6  PHI  796   0.556
   7  TEX  881   0.556
   8  TOR  746   0.549
   9  SEA  758   0.543

   Runs scored runs from ~700 to ~950; win percentage from ~0.54 to ~0.64. On a single axis, the second series would be a flat line glued to zero.

2. twinx() shares the x, splits the y

   ax2 = ax1.twinx() creates a second axes that shares ax1's x-axis but has its own y-axis on the right. Draw one series on each, and color each axis label to match its data so readers know which scale belongs to which.

   import matplotlib.pyplot as plt
   
   fig, ax1 = plt.subplots(figsize=(9, 5.5))
   ax1.bar(top["Abbr"], top["RS"], color="#B23A3A", alpha=0.85)      # left: runs
   ax1.set_ylabel("runs scored", color="#B23A3A")
   ax1.tick_params(axis="y", labelcolor="#B23A3A")
   
   ax2 = ax1.twinx()                                                 # share x, new right y
   ax2.plot(top["Abbr"], top["WinPct"], color="#2C5E8A", marker="o", lw=2)
   ax2.set_ylabel("win percentage", color="#2C5E8A")
   ax2.tick_params(axis="y", labelcolor="#2C5E8A")
   ax2.grid(False)                                                   # avoid clashing gridlines
   fig.savefig("dual_axis.png", dpi=144, bbox_inches="tight")

   

   One detail that saves headaches: turn the grid off on one axis. Two sets of gridlines at different scales never align and just add noise. Color-coding each axis to its series is the other must — without it, a dual axis is a puzzle.

3. The honesty caveat

   A dual axis has real power to deceive: because you choose each axis's range, you can make two series cross, diverge, or track together almost at will. Slide one axis up and a "relationship" appears that's pure axis choice. Use a dual axis only when the two series genuinely share an x-axis and you're showing levels, not implying a correlation. If your real claim is "these two move together", a scatter plot of one against the other is the honest chart — it can't be faked with axis limits. When in doubt, two stacked single-axis charts beat one clever dual axis.

Troubleshooting