Skip to main content

​
Overview

The generate_week_metrics() method calculates comprehensive weekly statistics for trip time, distance, and fare amounts. This provides a time-series view of taxi service patterns and enables week-over-week trend analysis.

​
How It Works

The method aggregates data by year_week (format: YYYY-WWW) to compute statistical measures and service counts:
main.py
def generate_week_metrics(self):
    # Calculate trip duration
    self.data['trip_time'] = self.data['tpep_dropoff_datetime'] - self.data['tpep_pickup_datetime']
    self.data['trip_time_in_seconds'] = self.data['trip_time'].dt.total_seconds()

    # Aggregate by week
    self.csv_df = self.data.groupby('year_week').agg(
        min_trip_time=('trip_time_in_seconds', 'min'),
        max_trip_time=('trip_time_in_seconds', 'max'),
        mean_trip_time=('trip_time_in_seconds', 'mean'),
        min_trip_distance=('trip_distance', 'min'),
        max_trip_distance=('trip_distance', 'max'),
        mean_trip_distance=('trip_distance', 'mean'),
        min_trip_amount=('total_amount', 'min'),
        max_trip_amount=('total_amount', 'max'),
        mean_trip_amount=('total_amount', 'mean'),
        total_services=('total_amount', 'count')
    ).reset_index()

    # Calculate week-over-week percentage change
    self.csv_df['percentage_variation'] = (
        self.csv_df['total_services'] - self.csv_df['total_services'].shift(1)
    ) / self.csv_df['total_services'].shift(1) * 100

​
Metrics Calculated

​
Trip Time Statistics

  • min_trip_time: Shortest trip duration in seconds
  • max_trip_time: Longest trip duration in seconds
  • mean_trip_time: Average trip duration in seconds
These metrics help identify:
  • Quick trips (likely short distances or off-peak hours)
  • Extended trips (long distances or heavy traffic)
  • Typical trip duration patterns by week

​
Trip Distance Statistics

  • min_trip_distance: Shortest trip distance in miles
  • max_trip_distance: Longest trip distance in miles
  • mean_trip_distance: Average trip distance in miles
Useful for understanding:
  • Service area coverage
  • Long-distance vs short-distance trip patterns
  • Distance trends over time

​
Trip Amount Statistics

  • min_trip_amount: Minimum fare in dollars
  • max_trip_amount: Maximum fare in dollars
  • mean_trip_amount: Average fare in dollars
Helps analyze:
  • Fare distribution
  • Revenue patterns
  • Pricing trends

​
Service Volume

  • total_services: Total number of completed trips in the week
Key metric for:
  • Demand analysis
  • Capacity planning
  • Trend identification

​
Week-over-Week Variation

  • percentage_variation: Percentage change in total services compared to the previous week
Calculated as: 
(current_week_services - previous_week_services) / previous_week_services × 100
This metric highlights:
  • Growth or decline in demand
  • Seasonal patterns
  • Impact of holidays or events
The first week in the dataset will have NaN for percentage_variation since there’s no previous week to compare against.

​
Data Grouping

Data is grouped by the year_week column, which is created during the add_more_columns() step:
main.py
self.data['year_dt'] = self.data['tpep_dropoff_datetime'].dt.year.astype(str)
self.data['week_dt'] = self.data['tpep_dropoff_datetime'].dt.isocalendar().week.astype(str).str.zfill(3)
self.data['year_week'] = self.data['year_dt'].str.cat(self.data['week_dt'], sep='-')
Weeks follow the ISO calendar standard:
  • Week 1 is the first week with a Thursday in the new year
  • Weeks run Monday through Sunday
  • Format: YYYY-WWW (e.g., 2022-001, 2022-052)

​
Output Format

The weekly metrics are stored in the csv_df DataFrame and exported as a pipe-delimited CSV file.

​
Column Order

year_week|min_trip_time|max_trip_time|mean_trip_time|min_trip_distance|max_trip_distance|mean_trip_distance|min_trip_amount|max_trip_amount|mean_trip_amount|total_services|percentage_variation

​
Example Output Data

year_week|min_trip_time|max_trip_time|mean_trip_time|min_trip_distance|max_trip_distance|mean_trip_distance|min_trip_amount|max_trip_amount|mean_trip_amount|total_services|percentage_variation
2022-001|60.0|7185.0|847.32|0.01|48.7|3.42|0.01|312.5|18.75|845623|
2022-002|60.0|7243.0|851.18|0.01|49.2|3.45|0.01|318.0|18.92|862341|1.98
2022-003|60.0|7156.0|839.64|0.01|47.8|3.38|0.01|308.25|18.53|878956|1.93
2022-004|60.0|7298.0|856.47|0.01|50.1|3.48|0.01|325.75|19.08|891234|1.40
2022-005|60.0|7087.0|832.91|0.01|46.9|3.35|0.01|302.5|18.35|854782|-4.09
2022-006|60.0|7412.0|868.53|0.01|51.3|3.52|0.01|332.0|19.45|897651|5.01
2022-007|60.0|7189.0|845.28|0.01|48.5|3.41|0.01|315.25|18.68|883429|-1.58
2022-008|60.0|7267.0|853.76|0.01|49.6|3.47|0.01|321.5|18.98|905817|2.54
2022-009|60.0|7134.0|837.92|0.01|47.4|3.36|0.01|306.75|18.46|869543|-4.00
2022-010|60.0|7345.0|862.15|0.01|50.8|3.51|0.01|328.0|19.23|916782|5.43
2022-011|60.0|7098.0|829.46|0.01|46.2|3.33|0.01|298.5|18.21|847291|-7.58
2022-012|60.0|7456.0|875.39|0.01|52.1|3.56|0.01|338.75|19.67|928456|9.58
2022-013|60.0|7203.0|848.64|0.01|48.9|3.43|0.01|317.25|18.79|891673|-3.96
All numeric values are rounded to 2 decimal places during the format_data() step.

​
Interpreting the Results

Look for patterns in the data:
  • Consistent positive percentage_variation: Growing demand
  • Consistent negative percentage_variation: Declining demand
  • Large fluctuations: Seasonal effects or special events
  • Stable mean values: Predictable service patterns

​
Example Analysis

From the sample data above:
  • Week 2022-005 shows a -4.09% drop in services (possibly a holiday week)
  • Week 2022-012 shows a +9.58% spike in services (possibly spring break)
  • Average trip time hovers around 840-875 seconds (~14-15 minutes)
  • Average trip distance is consistently 3.3-3.6 miles
  • Average fare ranges from $18-20

​
Anomaly Detection

Watch for:
  • Extreme min/max values that may indicate data quality issues
  • Sudden drops in total_services (system outages, weather events)
  • Unusual percentage variations (>20% week-over-week)

​
Use Cases

​
Demand Forecasting

Use historical percentage_variation patterns to predict future demand: 
import pandas as pd

# Load weekly metrics
df = pd.read_csv('processed_data.csv', sep='|')

# Calculate rolling average variation
df['rolling_avg_variation'] = df['percentage_variation'].rolling(window=4).mean()

# Forecast next week's services
last_services = df.iloc[-1]['total_services']
last_variation = df.iloc[-1]['rolling_avg_variation']
forecast = last_services * (1 + last_variation / 100)

​
Performance Benchmarking

Compare current week against historical averages: 
# Calculate overall averages
avg_trip_time = df['mean_trip_time'].mean()
avg_trip_distance = df['mean_trip_distance'].mean()
avg_trip_amount = df['mean_trip_amount'].mean()

# Compare current week
current_week = df.iloc[-1]
print(f"Time vs Avg: {(current_week['mean_trip_time'] / avg_trip_time - 1) * 100:.2f}%")
print(f"Distance vs Avg: {(current_week['mean_trip_distance'] / avg_trip_distance - 1) * 100:.2f}%")
print(f"Amount vs Avg: {(current_week['mean_trip_amount'] / avg_trip_amount - 1) * 100:.2f}%")

​
Visualization

Create time-series charts to visualize trends: 
import matplotlib.pyplot as plt

df = pd.read_csv('processed_data.csv', sep='|')

# Plot services over time
plt.figure(figsize=(12, 6))
plt.plot(df['year_week'], df['total_services'])
plt.xlabel('Week')
plt.ylabel('Total Services')
plt.title('Weekly Taxi Services')
plt.xticks(rotation=45)
plt.tight_layout()
plt.show()

​
Next Steps

Build docs developers (and LLMs) love

Get started for freeTalk to us