📘 Evaluating Demand Determinants for Yulu's Micro-Mobility Solutions

🧩 Business Problem

India’s largest micro-mobility platform, wants to understand what factors significantly impact the demand for electric bike rentals.

The dataset contains two years of rental data with variables such as season, weather, holiday, working day, temperature, humidity, and more. The company aims to identify statistically significant demand drivers to optimize:

• Fleet distribution
• Pricing strategies
• Operational planning
• Seasonal resource allocation

---

🎯 Objective

The goal of this project is to:

• Analyze the Yulu rental dataset across time, season, and environmental conditions.
• Perform hypothesis testing to determine which factors significantly affect rental demand.
• Identify patterns and trends in bike usage across seasons and working/holiday statuses.
• Translate analytical findings into business insights for Yulu’s operations team.

Column Profiling:

Column	Description
datetime	datetime
season	season (1: spring, 2: summer, 3: fall, 4: winter)
holiday	whether day is a holiday or not (extracted from http://dchr.dc.gov/page/holiday-schedule)
workingday	if day is neither weekend nor holiday is 1, otherwise is 0.
temp	temperature in Celsius
atemp	feeling temperature in Celsius
humidity	humidity
windspeed	wind speed
casual	count of casual users
registered	count of registered users
count	Total_riders count of total rental bikes including both casual and registered

weather:

Category	Details
1	Clear, Few clouds, partly cloudy, partly cloudy
2	Mist + Cloudy, Mist + Broken clouds, Mist + Few clouds, Mist
3	Light Snow, Light Rain + Thunderstorm + Scattered clouds, Light Rain + Scattered clouds
4	Heavy Rain + Ice Pallets + Thunderstorm + Mist, Snow + Fog

---

🛠️ Tasks Performed

• Loaded and inspected the dataset for schema, completeness, and variable types.

• Corrected inconsistent data types through type-casting (e.g., categorical variables).

• Conducted missing value and duplicate checks (none found).

• Explored all numerical and categorical variables using EDA.

• Visualized relationships using distribution plots, boxplots, and scatterplots.

• Analyzed seasonal and weather-driven variations in rentals.

• Performed Hypothesis Testing (t-tests, ANOVA) based on:

  • Season
  • Holiday vs. non-holiday
  • Working day vs. weekend
  • Weather condition

• Evaluated statistical significance (p-values, confidence levels).

• Interpreted results into actionable business recommendations.

---

🧠 Concepts Used

🔹 Exploratory Data Analysis (EDA)

• Distribution analysis
• Trend identification
• Outlier detection
• Categorical vs numerical comparisons

🔹 Statistical Hypothesis Testing

• t-tests
• ANOVA
• Confidence interval interpretation
• Understanding p-values

🔹 Data Preprocessing

• Type casting
• Feature understanding
• Categorizing temporal variables

🔹 Business Interpretation

• Understanding seasonal demand
• Effect of environmental variables
• Operational decision support

---

🔍 Findings & Observations

1. Data Quality

• No missing values
• No duplicates
• Two years of continuous rental data
• Season, holiday, workingday, and weather are clearly defined categorical variables

2. Seasonal Trends

• Bike rentals show visible seasonal fluctuations.
• Certain seasons demonstrate significantly higher rental counts.

3. Weather Conditions

• Adverse weather (rain, snow, storms) shows a noticeable drop in demand.
• Mild and clear weather conditions lead to peak rentals.

4. Working Day vs Weekend

• Working days have different rental patterns compared to weekends.
• Demand curves vary significantly across day types.

5. Holiday Impact

• Holidays show distinct usage patterns—often lower than working days.

---

📊 Hypothesis Testing Results (Summary)

Factor Tested	Test Used	Statistical Result	Business Interpretation
Season	ANOVA	Significant p-value	Season strongly influences rental demand
Holiday	t-test	Significant	Rentals differ on holidays vs normal days
Workingday	t-test	Significant	Weekdays vs weekends have different usage
Weather Situation	ANOVA	Significant	Weather conditions influence demand
Temperature / Humidity / Windspeed	Correlation + tests	Mixed significance	Some environmental variables correlate with demand

Conclusion:

Rental demand is statistically influenced by season, working day status, holiday status, and weather conditions.

---

🌦️ Demand Drivers & Seasonal Patterns

Based on EDA + hypothesis results:

High Demand Conditions

• ✔ Clear weather
• ✔ Moderate temperature
• ✔ Non-holiday weekdays
• ✔ Certain favorable seasons

Low Demand Conditions

• ✘ Holidays
• ✘ Adverse weather (rain, storms)
• ✘ High humidity or extreme temperatures

---

💡 Key Insights

• Demand is not uniform—seasonality plays a major role.
• Weather is a strong external factor affecting operational planning.
• Working days show a predictable commuter-based demand pattern.
• Holidays shift usage behavior → often leisure-driven.
• Environmental variables like humidity and temperature create non-linear effects.

---

📌 Recommendations

1️⃣ Optimize Fleet Allocation

• Increase the fleet during high-demand seasons.
• Reduce or redistribute during low-demand periods.

2️⃣ Weather-Aware Operations

• Plan fleet availability dynamically based on weather forecasts.
• Offer discounts in low-demand weather to maintain utilization.

3️⃣ Holiday & Weekend Strategy

• Weekend/holiday demand is different — promote leisure rides
• Offer weekend passes or tourist bundles

4️⃣ Pricing Strategy

• Dynamic pricing based on:

  • Peak seasons
  • High commuter periods
  • Weather-based supply-demand changes

5️⃣ Infrastructure Planning

• Place more bikes around office hubs for weekdays
• Add bikes near parks and public areas for weekends/holidays

---

🏁 Conclusion

This analysis demonstrates that Yulu’s bike rental demand is strongly driven by seasonality, weather, and day-type (working day vs holiday). Through hypothesis testing, we confirmed the statistical significance of these factors.