How to set a log axis scale in Matplotlib

Linear axes make equal numeric gaps look equal, which can bury patterns when a Matplotlib plot spans several powers of ten. A logarithmic axis spaces ticks by ratio, so growth from 100 to 1,000 uses the same visual interval as growth from 10,000 to 100,000.

The object-oriented Axes API keeps the scale change attached to one subplot. Set the vertical scale with ax.set_yscale(“log”), or set the horizontal scale with ax.set_xscale(“log”) when the x values span the wide range.

Use positive values on a normal logarithmic axis. Matplotlib uses base 10 by default for log scale, and data at zero or below needs a deliberate choice such as masking, clipping, or symlog before it is plotted.

1. Keep the values on the log-scaled axis positive.

   Values at zero or below cannot be positioned on a normal logarithmic scale. Mask or clip them deliberately, or use symlog when the plot must include values on both sides of zero.

2. Save the plotting script as log_axis_scale.py.

   log_axis_scale.py

   import matplotlib.pyplot as plt
    
   years = [2022, 2023, 2024, 2025, 2026]
   requests = [250, 1800, 12600, 94000, 710000]
    
   fig, ax = plt.subplots(layout="constrained")
   ax.plot(years, requests, marker="o")
   ax.set_yscale("log")
   ax.set_xticks(years)
   ax.set_xlabel("Year")
   ax.set_ylabel("Requests")
   ax.set_title("Requests by year")
   ax.grid(True, which="both", axis="y", color="0.9")
   fig.savefig("requests-log-axis.png", dpi=150)
    
   print(f"x scale: {ax.get_xscale()}")
   print(f"y scale: {ax.get_yscale()}")
   print("saved: requests-log-axis.png")

   ax.set_yscale(“log”) changes only the y-axis. Use ax.set_xscale(“log”) instead for a horizontal log axis, or use both calls when both dimensions need logarithmic spacing.

3. Run the script.

   $ python log_axis_scale.py
   x scale: linear
   y scale: log
   saved: requests-log-axis.png

4. Open the saved PNG and confirm the y-axis is spaced by powers of ten.

   Look for y scale: log in the output and powers-of-ten spacing on the y-axis in the image.