Visualizing Real-World Bike Rental Data with Python

I recently worked on a dataset tracking daily bike rentals over time — a perfect use case to test out how Python visualizations can turn raw time series into readable insights.

The project started with chaos: jagged lines, unreadable axes, and unclear stories. But with a few critical techniques — rolling averages, resampling, and Matplotlib’s hidden tricks — the data began to speak clearly.

Here’s a breakdown of how that happened.

The Dataset

The dataset included:

• Daily rental counts

• Timestamps

• Weather conditions

It was messy enough to require real cleaning — and that made the learning deeper.

1) Turning That Chaos Into Clarity (Step by Step)

Here’s the exact process I followed to go from that mess to something readable:

1. Convert ‘Date’ column to datetime:

    df['Date'] = pd.to_datetime(df['Date'])
   

2. Set it as index to enable resampling:

    df.set_index('Date', inplace=True)
   

3. Handle missing values (if any):

    df.dropna(inplace=True)
   

4. Smooth trends using a rolling average:

    df['7D_Rolling'] = df['Rental Count'].rolling(window=7).mean()
   

5. Resample to monthly average:

    monthly_avg = df['Rental Count'].resample('M').mean()
   

6. Polish with figure sizing + axis formatting:

    import matplotlib.dates as mdates
    ax = df['7D_Rolling'].plot(figsize=(12, 6), label='7-day Rolling Avg')
    ax.xaxis.set_major_locator(mdates.MonthLocator())
    ax.xaxis.set_major_formatter(mdates.DateFormatter('%b'))
    plt.legend()
   

---

2) Average Rentals by Weekday

Once the time series was clean, I got curious:
Do certain weekdays consistently get more rentals?

Hypothetical Problem:

The dataset had Date as index but no Day column. I couldn’t directly group by day names.

Fix:

I extracted day names from the datetime index:

df['Day'] = df.index.day_name()
weekday_avg = df.groupby('Day')['Rental Count'].mean().sort_values()

3) Finding the Most Popular Hours

Then I wanted to know:
At what time of day are rentals most frequent?

Hypothetical Problem:

The data had timestamps, but hours weren’t separate. Plotting directly would group by full datetime — unreadable.

Fix:

I extracted the hour from the timestamp:

df['Hour'] = df.index.hour
hourly_avg = df.groupby('Hour')['Rental Count'].mean()

Final Thoughts

This wasn’t just a visualization project — it was about designing clarity in the middle of chaos.

• The raw data looked meaningless until I smoothed, grouped, and reframed it

• Every tiny formatting step made the insight more obvious

• And when I zoomed in by hour and out by month, the human behavior in the data started showing

No flashy dashboards. Just thoughtful plots that told real stories.

---