Dates and Trends

A single number means little. Revenue of a million dollars. Is that good? Compared to what? The story is almost always in the change. Pandas has wonderful tools for dates, so let's use them.

• One number means little
• Compared to what?
• The story is in the change

Weekdays and growth

When a column is a real date, the dt accessor gives you the weekday, the month, and more. We find the busiest weekdays, then sum revenue by month, and compare the last year with the first. Run it. Weekends are the strongest. And the business has grown over the year. The founder can use this for planning staff.

import pandas as pd

sales = pd.read_csv("sales.csv", parse_dates=["date"])

sales["weekday"] = sales["date"].dt.day_name()
sales["month"] = sales["date"].dt.to_period("M")

print(sales.groupby("weekday")["revenue"].sum().sort_values().round(0).tail(3).to_string())
monthly = sales.groupby("month")["revenue"].sum()
print(monthly.tail(3).round(0).to_string())
print("Growth over the last year:", f"{monthly.iloc[-12:].sum() / monthly.iloc[:12].sum() - 1:.1%}")

weekday
Wednesday     869304.0
Sunday       1067971.0
Saturday     1074394.0
month
2025-10    304887.0
2025-11    328751.0
2025-12    356262.0
Freq: M
Growth over the last year: 15.2%

Rolling averages

Daily numbers are noisy. A rolling average smooths them out. Rolling twenty eight averages the last four weeks, so whole weeks are included, which cancels the weekend effect. Percent change compares with a week ago. Run it. Now we can see the true level, and the direction. Always compare a weekday with the same weekday, or you'll just measure the weekend.

import pandas as pd

sales = pd.read_csv("sales.csv", parse_dates=["date"])
daily = sales.groupby("date")["revenue"].sum()

trend = daily.rolling(28).mean()
change = daily.pct_change(7)

print(daily.tail(2).round(0).to_string())
print("28 day average:", round(trend.iloc[-1]))
print("Change on last week:", f"{change.iloc[-1]:+.1%}")
print("Best day so far:", daily.idxmax().date(), round(daily.max()))

date
2025-12-30    10620.0
2025-12-31    11017.0
28 day average: 11608
Change on last week: +2.0%
Best day so far: 2025-12-28 13883

Summary