Your first plot

Here is the whole thing, and then we'll take it apart line by line.

fig, ax = plt.subplots(figsize=(6, 3.5))
ax.plot(days, delhi, label="Delhi")
ax.plot(days, mumbai, label="Mumbai")
ax.set_xlabel("day")
ax.set_ylabel("cups")
ax.set_title("Cups sold this week")
ax.legend()
ax.grid(True, alpha=0.3)

Notice the proportion: two lines put data on the chart and six make it legible. That ratio is normal, and it's the part tutorials skip.

The frame

plt.subplots() is the one you should reach for by default. It creates a Figure with one Axes inside and returns both. figsize is in inches — width first — which tells you roughly how old this library is. Multiply by dpi for pixels.

The data

ax.plot(days, delhi)   # x first, then y

With one argument, ax.plot(delhi), matplotlib uses 0, 1, 2… as x — handy, and an easy way to draw the wrong chart without noticing. Pass both.

Call plot again on the same Axes and it draws on top, taking the next colour from the property cycle. You don't pick colours unless you want to; the cycle keeps series distinguishable for you. Common overrides:

ax.plot(days, delhi, color="#66aaf9", linewidth=2,
        linestyle="--", marker="o", alpha=0.8)

There's also the compact format string — ax.plot(x, y, "r--o") for a red dashed line with circle markers. It's terse and it appears everywhere online, but the keyword form is the one to write: it's readable a month later.

Saying what it is

set_xlabel, set_ylabel, set_title. Nothing adds these for you, and an unlabelled axis is a chart that only makes sense to the person who made it, for about twenty minutes.

Units belong in the label: "revenue (₹ thousands)", not "revenue". And a title is better spent on the finding than the columns — "Weekend trade is up 40%" beats "cups by day", because the axis labels already said what the columns are.

The legend

ax.legend() builds itself from the label= you passed to each plotting call. No labels, no legend — you'll get an empty box and a warning instead. Two arguments earn their keep:

ax.legend(loc="upper left")     # or "best", the default
ax.legend(frameon=False)         # no box — usually cleaner

With a single series, skip the legend entirely and put the name in the title. A legend explaining one line is furniture.

The grid

ax.grid(True, alpha=0.3). Faint on purpose — it should help the eye carry a value across to the axis without competing with the data. A grid at full strength is a cage.

Checking what you got

Every setter has a matching getter, which is how you verify a chart without squinting:

ax.get_title()      ax.get_xlabel()     ax.get_ylim()
len(ax.lines)       ax.lines[0].get_color()