Matplotlib Tutorial (Part 8): Plotting Time Series Data

Time-series plotting in Matplotlib hinges on treating dates as real datetime objects—not plain strings—and then using Matplotlib’s date formatting tools to make the x-axis readable. The tutorial starts with a simple seven-day dataset created directly in Python, plots it with `plt.plot_date`, and then adjusts the line style so points connect into a continuous trend. From there, it focuses on two practical upgrades that matter in real charts: automatic rotation/alignment of date tick labels and custom date label formats (like “May 24” instead of the default year-month-day).

The first example builds a list of consecutive `datetime` values and a matching list of y-values. Plotting is straightforward: pass dates as the x-axis and the numeric series as the y-axis. The initial output may show markers rather than a line, but setting `linestyle='solid'` produces a connected time series; markers can be disabled with `marker=None` if desired. To improve readability, the chart uses Matplotlib’s figure-level auto-formatting for dates—calling `autofmt_xdate()` on the current figure—so tick labels rotate and spread out instead of bunching together.

Next comes the key formatting step: changing how dates appear on the x-axis. The tutorial imports `matplotlib.dates` as `MPL dates` and uses `matplotlib.dates.DateFormatter` with a format string compatible with Python’s `strftime` codes. It then applies that formatter to the x-axis via `ax.xaxis.set_major_formatter(...)` after retrieving the current axis. The example uses abbreviated month, day, and four-digit year (e.g., `%b %d %Y`), resulting in labels like “May 24” rather than full numeric date strings.

The real-world section loads Bitcoin OHLCV data from a CSV using pandas. The CSV includes a `date` column plus columns such as open, high, low, close, adjusted close, and volume. The code initially plots `price_date` (from the CSV’s `date` column) against `price_close` using `plot_date`, but the x-axis renders incorrectly because the dates are still strings. A deliberate out-of-order date row demonstrates the problem: the chart places “May 17” at the end rather than at the correct chronological position.

The fix is twofold. First, convert the `date` column to pandas datetime objects with `pd.to_datetime(...)`, replacing the string values. Second, sort the data by date so the time series is ordered correctly; the tutorial uses `sort_values` (with `inplace=True`) to reorder the DataFrame without needing to reassign it. After these changes, Matplotlib receives properly typed datetimes, and the x-axis behaves as expected.

The session closes by previewing a follow-up on real-time plotting—useful for monitoring data that updates continuously from APIs or sensors—while reinforcing that the core workflow for time series is: parse dates correctly, sort chronologically, plot with `plot_date`, and then format the x-axis for human readability.