This project explores the Bike Sharing Demand dataset to understand factors that influence bike rental demand in Washington D.C. The bike sharing systems allow to people rent a bike from a one location and return it to a different place on an as-needed basis.
This section performs the initial loading of the dataset and a first general inspection to understand the structure and available attributes.
- Load and explore the dataset
- Analyze hourly, daily, and seasonal bike usage
- Investigate the impact of weather and temperature
- Visualize distributions and detect outliers
- Evaluate correlations between features
- Python
- Pandas
- NumPy
- Seaborn
- Matplotlib
- Jupyter Notebook
- Github
# Import libraries
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
pd.set_option('display.float_format', '{:.2f}'.format)the data provide hourly rental data spanning two years and combine usage patterns with weather data to forescast bike rental demand in Washington D.C.
Dataset used: bike-sharing-demand.csv Including attributes as:
datetime: hourly date + timestamp
season: (1) = spring, (2) = summer, (3) = fall, (4) = winter
holiday: whether the day is considered a holiday
workingday: whether the day is neither a weekend nor holiday
weather: (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
temp: real temperature in Celsius
atemp: "feels like" temperature in Celsius
humidity: relative humidity
windspeed: wind speed
casual: number of non-registered user rentals initiated
registered: number of registered user rentals initiated
count: number of total rentals
# Load the dataset
bike_sharing_demand_df = pd.read_csv("intro_to_github_dir/project/Bike_Sharing_Demand/bike-sharing-demand.csv")# Display the first few rows to confirm loading
bike_sharing_demand_df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| datetime | season | holiday | workingday | weather | temp | atemp | humidity | windspeed | casual | registered | count | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 01/01/2011 00:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 81 | 0.00 | 3 | 13 | 16 |
| 1 | 01/01/2011 01:00 | 1 | 0 | 0 | 1 | 9.02 | 13.63 | 80 | 0.00 | 8 | 32 | 40 |
| 2 | 01/01/2011 02:00 | 1 | 0 | 0 | 1 | 9.02 | 13.63 | 80 | 0.00 | 5 | 27 | 32 |
| 3 | 01/01/2011 03:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 75 | 0.00 | 3 | 10 | 13 |
| 4 | 01/01/2011 04:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 75 | 0.00 | 0 | 1 | 1 |
Columns, datatype and null vals. Broader picture of the dataset.
- The dataframe has 10886 data point (registers), with 12 columns which are attributes. 11 atributes to describe weather and and bike usage patterns.
- Dtype: there is not null values in any column. int64: integers, float64: decimal numb and object(datetime): text or dates as a string
- memory usage in the comp, aprox 1 megabyte
- int64 works like a categorical data because describe numerical codification labels
bike_sharing_demand_df.info()<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10886 entries, 0 to 10885
Data columns (total 12 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 datetime 10886 non-null object
1 season 10886 non-null int64
2 holiday 10886 non-null int64
3 workingday 10886 non-null int64
4 weather 10886 non-null int64
5 temp 10886 non-null float64
6 atemp 10886 non-null float64
7 humidity 10886 non-null int64
8 windspeed 10886 non-null float64
9 casual 10886 non-null int64
10 registered 10886 non-null int64
11 count 10886 non-null int64
dtypes: float64(3), int64(8), object(1)
memory usage: 1020.7+ KB
To know attributes and verify column names
bike_sharing_demand_df.columnsIndex(['datetime', 'season', 'holiday', 'workingday', 'weather', 'temp',
'atemp', 'humidity', 'windspeed', 'casual', 'registered', 'count'],
dtype='object')
bike_sharing_demand_df.describe<bound method NDFrame.describe of datetime season holiday workingday weather temp atemp \
0 01/01/2011 00:00 1 0 0 1 9.84 14.39
1 01/01/2011 01:00 1 0 0 1 9.02 13.63
2 01/01/2011 02:00 1 0 0 1 9.02 13.63
3 01/01/2011 03:00 1 0 0 1 9.84 14.39
4 01/01/2011 04:00 1 0 0 1 9.84 14.39
... ... ... ... ... ... ... ...
10881 19/12/2012 19:00 4 0 1 1 15.58 19.70
10882 19/12/2012 20:00 4 0 1 1 14.76 17.43
10883 19/12/2012 21:00 4 0 1 1 13.94 15.91
10884 19/12/2012 22:00 4 0 1 1 13.94 17.43
10885 19/12/2012 23:00 4 0 1 1 13.12 16.66
humidity windspeed casual registered count
0 81 0.00 3 13 16
1 80 0.00 8 32 40
2 80 0.00 5 27 32
3 75 0.00 3 10 13
4 75 0.00 0 1 1
... ... ... ... ... ...
10881 50 26.00 7 329 336
10882 57 15.00 10 231 241
10883 61 15.00 4 164 168
10884 61 6.00 12 117 129
10885 66 9.00 4 84 88
[10886 rows x 12 columns]>
print("season:", bike_sharing_demand_df['season'].unique())
print("holiday:", bike_sharing_demand_df['holiday'].unique())
print("workingday:", bike_sharing_demand_df['workingday'].unique())
print("weather:", bike_sharing_demand_df['weather'].unique())season: [1 2 3 4]
holiday: [0 1]
workingday: [0 1]
weather: [1 2 3 4]
Since the result is an empty DataFrame, that confirms there are no rows with any missing values.
bike_sharing_demand_df.isnull().sum()datetime 0
season 0
holiday 0
workingday 0
weather 0
temp 0
atemp 0
humidity 0
windspeed 0
casual 0
registered 0
count 0
dtype: int64
null_val = bike_sharing_demand_df[bike_sharing_demand_df.isna().any(axis=1)]
print(null_val)Empty DataFrame
Columns: [datetime, season, holiday, workingday, weather, temp, atemp, humidity, windspeed, casual, registered, count]
Index: []
duplicados = bike_sharing_demand_df.duplicated().sum()
print(f"Duduplicados encontrados: {duplicados}")Duduplicados encontrados: 0
As we need information about the rental hours of the bicycles, we need to extract this information from the datetime column, which is of type object.
bike_sharing_demand_df['datetime'] = pd.to_datetime(bike_sharing_demand_df['datetime'], dayfirst=True)
print(bike_sharing_demand_df["datetime"].dtype)datetime64[ns]
#Convert Datetime to numeric column
bike_sharing_demand_df['datetime'] = pd.to_datetime(bike_sharing_demand_df['datetime'])
# Temporal analysis
bike_sharing_demand_df['hour'] = bike_sharing_demand_df['datetime'].dt.hour # 0-23 (int)
bike_sharing_demand_df['day'] = bike_sharing_demand_df['datetime'].dt.day # 1-31 (int)
bike_sharing_demand_df['month'] = bike_sharing_demand_df['datetime'].dt.month # 1-12 (int)
bike_sharing_demand_df['year'] = bike_sharing_demand_df['datetime'].dt.year # 2011, 2012 (int)
bike_sharing_demand_df['weekday'] = bike_sharing_demand_df['datetime'].dt.dayofweek # 0-6 (int, 0=Mon)
# Dtypes
print(bike_sharing_demand_df[['hour', 'day', 'month', 'year', 'weekday']].dtypes)hour int32
day int32
month int32
year int32
weekday int32
dtype: object
# Null values in datetime
print(bike_sharing_demand_df["datetime"].isnull().sum())
print(bike_sharing_demand_df["datetime"].unique()[10])0
2011-01-01 10:00:00
bike_sharing_demand_df['datetime'] = pd.to_datetime(bike_sharing_demand_df['datetime'], errors='coerce')
# This converts problematic values to NaT (Not a Time).bike_sharing_demand_df.head()
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
</style>
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
| datetime | season | holiday | workingday | weather | temp | atemp | humidity | windspeed | casual | registered | count | hour | day | month | year | weekday | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 2011-01-01 00:00:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 81 | 0.00 | 3 | 13 | 16 | 0 | 1 | 1 | 2011 | 5 |
| 1 | 2011-01-01 01:00:00 | 1 | 0 | 0 | 1 | 9.02 | 13.63 | 80 | 0.00 | 8 | 32 | 40 | 1 | 1 | 1 | 2011 | 5 |
| 2 | 2011-01-01 02:00:00 | 1 | 0 | 0 | 1 | 9.02 | 13.63 | 80 | 0.00 | 5 | 27 | 32 | 2 | 1 | 1 | 2011 | 5 |
| 3 | 2011-01-01 03:00:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 75 | 0.00 | 3 | 10 | 13 | 3 | 1 | 1 | 2011 | 5 |
| 4 | 2011-01-01 04:00:00 | 1 | 0 | 0 | 1 | 9.84 | 14.39 | 75 | 0.00 | 0 | 1 | 1 | 4 | 1 | 1 | 2011 | 5 |
print(bike_sharing_demand_df.info())
print(bike_sharing_demand_df.describe())<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10886 entries, 0 to 10885
Data columns (total 17 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 datetime 10886 non-null datetime64[ns]
1 season 10886 non-null int64
2 holiday 10886 non-null int64
3 workingday 10886 non-null int64
4 weather 10886 non-null int64
5 temp 10886 non-null float64
6 atemp 10886 non-null float64
7 humidity 10886 non-null int64
8 windspeed 10886 non-null float64
9 casual 10886 non-null int64
10 registered 10886 non-null int64
11 count 10886 non-null int64
12 hour 10886 non-null int32
13 day 10886 non-null int32
14 month 10886 non-null int32
15 year 10886 non-null int32
16 weekday 10886 non-null int32
dtypes: datetime64[ns](1), float64(3), int32(5), int64(8)
memory usage: 1.2 MB
None
datetime season holiday workingday weather \
count 10886 10886.00 10886.00 10886.00 10886.00
mean 2011-12-27 05:56:22.399411968 2.51 0.03 0.68 1.42
min 2011-01-01 00:00:00 1.00 0.00 0.00 1.00
25% 2011-07-02 07:15:00 2.00 0.00 0.00 1.00
50% 2012-01-01 20:30:00 3.00 0.00 1.00 1.00
75% 2012-07-01 12:45:00 4.00 0.00 1.00 2.00
max 2012-12-19 23:00:00 4.00 1.00 1.00 4.00
std NaN 1.12 0.17 0.47 0.63
temp atemp humidity windspeed casual registered count \
count 10886.00 10886.00 10886.00 10886.00 10886.00 10886.00 10886.00
mean 20.23 23.66 61.89 12.80 36.02 155.55 191.57
min 0.82 0.76 0.00 0.00 0.00 0.00 1.00
25% 13.94 16.66 47.00 7.00 4.00 36.00 42.00
50% 20.50 24.24 62.00 13.00 17.00 118.00 145.00
75% 26.24 31.06 77.00 17.00 49.00 222.00 284.00
max 41.00 45.45 100.00 57.00 367.00 886.00 977.00
std 7.79 8.47 19.25 8.16 49.96 151.04 181.14
hour day month year weekday
count 10886.00 10886.00 10886.00 10886.00 10886.00
mean 11.54 9.99 6.52 2011.50 3.01
min 0.00 1.00 1.00 2011.00 0.00
25% 6.00 5.00 4.00 2011.00 1.00
50% 12.00 10.00 7.00 2012.00 3.00
75% 18.00 15.00 10.00 2012.00 5.00
max 23.00 19.00 12.00 2012.00 6.00
std 6.92 5.48 3.44 0.50 2.00
Acording with the time pattern, is observed that the highest demand for bike rentals occurs during working hours, with the peaks between 7-9 and 16-19 h. It´s noted that durent the night-time period, demanc is very low.
# Rented bikes (average) per Hour
hourly_pattern = bike_sharing_demand_df.groupby('hour')['count'].mean()
hourly_pattern.plot(kind='bar', figsize=(12, 6), title='Rented bikes (average) per Hour')
plt.xlabel('Hour of the day')
plt.ylabel('Rented bikes (average)')
plt.xticks(rotation=0)
plt.show()
# in numbers
print("=== TIME PATTERN ===")
print(hourly_pattern.round(2))
=== TIME PATTERN ===
hour
0 55.14
1 33.86
2 22.90
3 11.76
4 6.41
5 19.77
6 76.26
7 213.12
8 362.77
9 221.78
10 175.09
11 210.67
12 256.51
13 257.79
14 243.44
15 254.30
16 316.37
17 468.77
18 430.86
19 315.28
20 228.52
21 173.37
22 133.58
23 89.51
Name: count, dtype: float64
# Creating labels for week days
day_labels = ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"]
weekly_pattern = bike_sharing_demand_df.groupby('weekday')['count'].mean()
# Graphic
weekly_pattern.plot(kind='bar', figsize=(10, 6), title='Rented bikes (average) per day of week')
plt.xlabel('Weekday')
plt.ylabel('Rented bikes (average)')
plt.xticks(range(7), day_labels, rotation=0)
plt.show()
Since the analysis per day does not show a clear pattern, an analysis pero hour per day would be clarified the behaviour of the rental system.
# Creating labels for week days
day_labels = ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"]
weekly_pattern = bike_sharing_demand_df.groupby('weekday')['count'].mean()
# Creating time pattern per day of the week
hourly_weekly_pattern = bike_sharing_demand_df.groupby(['hour', 'weekday'])['count'].mean().reset_index()
# Labels
day_name = {0: 'Mon', 1: 'Tue', 2: 'Wed', 3: 'Thu',
4: 'Fri', 5: 'Sat', 6: 'Sun'}
hourly_weekly_pattern['day_name'] = hourly_weekly_pattern['weekday'].map(day_name)
# Style of the graphic
plt.style.use('seaborn-v0_8')
fig, ax = plt.subplots(figsize=(14, 8))
# Colours
colors = ['#1f77b4', '#ff7f0e', '#2ca02c', '#d62728', '#9467bd', '#17becf', '#e377c2']
weekend_colors = ['#17becf', '#e377c2']
# lines graphic
for i, day in enumerate(range(7)):
day_data = hourly_weekly_pattern[hourly_weekly_pattern['weekday'] == day]
# for weekend
linewidth = 3 if day >= 5 else 2
linestyle = '-' if day < 5 else '--'
plt.plot(day_data['hour'], day_data['count'],
marker='o', markersize=4, linewidth=linewidth, linestyle=linestyle,
label=day_name[day], color=colors[i], alpha=0.8)
plt.xlabel('Hour', fontsize=14, fontweight='bold')
plt.ylabel('Rented Bikes (average)', fontsize=14, fontweight='bold')
plt.title('Rented Bikes Demand per hour and day',
fontsize=16, fontweight='bold', pad=20)
plt.xlim(-0.5, 23.5)
plt.xticks(range(0, 24, 2))
plt.grid(True, alpha=0.3)
legend = plt.legend(bbox_to_anchor=(1.05, 1), loc='upper left',
frameon=True, fancybox=True, shadow=True)
legend.get_frame().set_facecolor('white')
legend.get_frame().set_alpha(0.9)
plt.tight_layout()
plt.show()
# Análisis estadístico adicional
print("=== Analysis per day ===")
print("\n1. Average per Day:")
daily_averages = bike_sharing_demand_df.groupby('weekday')['count'].mean()
for day, avg in daily_averages.items():
print(f" {day_name[day]}: {avg:.1f} bicicletas/hora")
print("\n2. Peak hours per Day:")
for day in range(7):
day_data = hourly_weekly_pattern[hourly_weekly_pattern['weekday'] == day]
peak_hour = day_data.loc[day_data['count'].idxmax(), 'hour']
peak_demand = day_data['count'].max()
print(f" {day_name[day]}: Hora {peak_hour}:00 ({peak_demand:.1f} bicicletas)")
print("\n3. Hour with less demand per day:")
for day in range(7):
day_data = hourly_weekly_pattern[hourly_weekly_pattern['weekday'] == day]
low_hour = day_data.loc[day_data['count'].idxmin(), 'hour']
low_demand = day_data['count'].min()
print(f" {day_name[day]}: Hour {low_hour}:00 ({low_demand:.1f} bikes)")
# Gráfico adicional: Heatmap de demanda
print("\n=== HEATMAP DEMAND ===")
fig, ax = plt.subplots(figsize=(12, 6))
# Matrix
heatmap_data = hourly_weekly_pattern.pivot(index='day_name', columns='hour', values='count')
day_order = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
heatmap_data = heatmap_data.reindex(day_order)
# heatmap
sns.heatmap(heatmap_data, annot=False, cmap='YlOrRd', fmt='.0f',
cbar_kws={'label': 'Rented Bikes (average)'})
plt.title('Heatmap: Rented Bikes Demand per hour and day', fontsize=14, fontweight='bold')
plt.xlabel('Hour', fontsize=14)
plt.ylabel('Day', fontsize=14)
plt.tight_layout()
plt.show()
=== Analysis per day ===
1. Average per Day:
Mon: 190.4 bicicletas/hora
Tue: 189.7 bicicletas/hora
Wed: 188.4 bicicletas/hora
Thu: 197.3 bicicletas/hora
Fri: 197.8 bicicletas/hora
Sat: 196.7 bicicletas/hora
Sun: 180.8 bicicletas/hora
2. Peak hours per Day:
Mon: Hora 17:00 (521.4 bicicletas)
Tue: Hora 17:00 (544.2 bicicletas)
Wed: Hora 17:00 (509.3 bicicletas)
Thu: Hora 17:00 (536.6 bicicletas)
Fri: Hora 17:00 (502.0 bicicletas)
Sat: Hora 13:00 (402.9 bicicletas)
Sun: Hora 13:00 (384.8 bicicletas)
3. Hour with less demand per day:
Mon: Hour 3:00 (5.7 bikes)
Tue: Hour 3:00 (4.0 bikes)
Wed: Hour 4:00 (4.6 bikes)
Thu: Hour 3:00 (4.9 bikes)
Fri: Hour 4:00 (5.9 bikes)
Sat: Hour 4:00 (7.7 bikes)
Sun: Hour 5:00 (9.5 bikes)
=== HEATMAP DEMAND ===
# by month
monthly_pattern = bike_sharing_demand_df.groupby('month')['count'].mean()
monthly_pattern.plot(kind='line', marker='o', figsize=(12, 6), title='Rented bikes (average) per month')
plt.xlabel('Month')
plt.ylabel('Rented bikes (average)')
plt.xticks(range(1, 13))
plt.grid(True)
plt.show()
correlation between all the attributes.
#Only numeric columns
numeric_cols = ['count', 'casual', 'registered', 'temp', 'atemp', 'humidity',
'windspeed', 'hour', 'day', 'month', 'year', 'weekday', 'weather']
#Correlation matrix
corr_matrix = bike_sharing_demand_df[numeric_cols].corr()
# show correlations with "count" sorted to see which have more weight correlation
print("=== Correlation - Total Demand (count) ===")
count_correlations = corr_matrix['count'].abs().sort_values(ascending=False)
print(count_correlations)=== Correlation - Total Demand (count) ===
count 1.00
registered 0.97
casual 0.69
hour 0.40
temp 0.39
atemp 0.39
humidity 0.32
year 0.26
month 0.17
weather 0.13
windspeed 0.10
day 0.02
weekday 0.00
Name: count, dtype: float64
#Correlation Matrix - Heatmap
#correlation between all the attributes.
#Only numeric columns
numeric_cols = ['count', 'casual', 'registered', 'temp', 'atemp', 'humidity',
'windspeed', 'hour', 'day', 'month', 'year', 'weekday', 'weather']
#Correlation matrix
corr_matrix = bike_sharing_demand_df[numeric_cols].corr()
#display the matrix as a heatmap
plt.figure(figsize=(10, 8))
sns.heatmap(corr_matrix, annot=True, cmap="coolwarm", center=0, square=True, fmt=".2f")
plt.title("Correlation Matrix - Bike sharing Demand")
plt.show()
print("\n=== TOP CORRELATIONS DEMAND ===")
print(f"Temperatura: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['temp']):.3f}")
print(f"Temperatura aparente: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['atemp']):.3f}")
print(f"Humedad: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['humidity']):.3f}")
print(f"Velocidad viento: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['windspeed']):.3f}")
print(f"Hora del día: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['hour']):.3f}")
print(f"Mes del año: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['month']):.3f}")
print(f"Condición climática: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['weather']):.3f}")=== TOP CORRELATIONS DEMAND ===
Temperatura: 0.394
Temperatura aparente: 0.390
Humedad: -0.317
Velocidad viento: 0.101
Hora del día: 0.401
Mes del año: 0.167
Condición climática: -0.129
# Converting in Categorical
# keep original df (numeric for corr)
# season (1,2,3,4) y weather (1,2,3,4) continue exist
# new columns - Season
season_mapping = {
1: 'Spring', 2: 'Summer', 3: 'Autumn', 4: 'Winter'
}
weather_mapping = {
1: 'Clear', 2: 'Fog', 3: 'Light Rain', 4: 'Heavy Rain'
}
bike_sharing_demand_df['season_label'] = bike_sharing_demand_df['season'].map(season_mapping)
bike_sharing_demand_df['weather_label'] = bike_sharing_demand_df['weather'].map(weather_mapping)
# confirmation of the both df dtype
print("Numeric attributes (for corralations):")
print(f"season: {bike_sharing_demand_df['season'].dtype}") # int64
print(f"weather: {bike_sharing_demand_df['weather'].dtype}") # int64
print("\nCategoric (for interpretation):")
print(f"season_label: {bike_sharing_demand_df['season_label'].dtype}") # object
print(f"weather_label: {bike_sharing_demand_df['weather_label'].dtype}") # objectNumeric attributes (for corralations):
season: int64
weather: int64
Categoric (for interpretation):
season_label: object
weather_label: object
# Correlations still work
print(f"Correlación season: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['season']):.3f}")
print(f"Correlación weather: {bike_sharing_demand_df['count'].corr(bike_sharing_demand_df['weather']):.3f}")Correlación season: 0.163
Correlación weather: -0.129
# More clear for reading
print(bike_sharing_demand_df.groupby('season_label')['count'].mean())
print(bike_sharing_demand_df.groupby('weather_label')['count'].mean())season_label
Autumn 234.42
Spring 116.34
Summer 215.25
Winter 198.99
Name: count, dtype: float64
weather_label
Clear 205.24
Fog 178.96
Heavy Rain 164.00
Light Rain 118.85
Name: count, dtype: float64
# Specific correlations using season and weather
print("=== CORRELATIONS WITH SEASON ===")
season_corr = bike_sharing_demand_df[['count', 'casual', 'registered', 'temp', 'atemp', 'humidity',
'windspeed', 'hour', 'month']].corrwith(bike_sharing_demand_df['season'])
print(season_corr.abs().sort_values(ascending=False))
print("\n=== CORRELATIONS WITH WEATHER ===")
weather_corr = bike_sharing_demand_df[['count', 'casual', 'registered', 'temp', 'atemp', 'humidity',
'windspeed', 'hour', 'month']].corrwith(bike_sharing_demand_df['weather'])
print(weather_corr.abs().sort_values(ascending=False))=== CORRELATIONS WITH SEASON ===
month 0.97
atemp 0.26
temp 0.26
humidity 0.19
registered 0.16
count 0.16
windspeed 0.15
casual 0.10
hour 0.01
dtype: float64
=== CORRELATIONS WITH WEATHER ===
humidity 0.41
casual 0.14
count 0.13
registered 0.11
atemp 0.06
temp 0.06
hour 0.02
month 0.01
windspeed 0.01
dtype: float64
# Heatmap with the most relevant correlated attributes
key_vars = ['count', 'temp', 'humidity', 'season', 'weather', 'hour']
plt.figure(figsize=(8, 6))
sns.heatmap(bike_sharing_demand_df[key_vars].corr(), annot=True, cmap='coolwarm', center=0,
square=True, fmt='.2f', cbar_kws={'label': 'Correlation'})
plt.title('Relevant Correlations: Demand (count), weather, and season')
plt.show()
# cross table: Season vs Weather with demand average
season_weather_demand = bike_sharing_demand_df.groupby(['season_label', 'weather_label'])['count'].mean().unstack()
print("=== DEMAND AVERAGE: SEASON vs WEATHER CONDITIONS ===")
print(season_weather_demand.round(2))
# Heatmap de demanda por estación y clima
plt.figure(figsize=(10, 6))
sns.heatmap(season_weather_demand, annot=True, cmap='YlOrRd', fmt='.0f',
cbar_kws={'label': 'Promedio de Rentals'})
plt.title('Demanda de Bicicletas: Estación vs Condición Climática')
plt.xlabel('Condición Climática')
plt.ylabel('Estación del Año')
plt.show()=== DEMAND AVERAGE: SEASON vs WEATHER CONDITIONS ===
weather_label Clear Fog Heavy Rain Light Rain
season_label
Autumn 243.58 230.77 NaN 156.58
Spring 126.78 106.86 164.00 61.23
Summer 236.73 189.52 NaN 123.91
Winter 209.51 194.78 NaN 134.47
# Ver qué tan común es cada combinación
season_weather_count = bike_sharing_demand_df.groupby(['season_label', 'weather_label']).size().unstack(fill_value=0)
plt.figure(figsize=(10, 6))
sns.heatmap(season_weather_count, annot=True, cmap='Purples', fmt='d',
cbar_kws={'label': 'Número de Horas'})
plt.title('Frecuencia de Condiciones Climáticas por Estación')
plt.xlabel('Condición Climática')
plt.ylabel('Estación del Año')
plt.show()
# Crear subplots para season y weather
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(15, 6))
# Histograma de Season (usando etiquetas)
season_counts = bike_sharing_demand_df['season_label'].value_counts()
season_counts.plot(kind='bar', ax=ax1, color=['lightgreen', 'orange', 'brown', 'lightblue'])
ax1.set_title('Distribución de Registros por Estación')
ax1.set_xlabel('Estación del Año')
ax1.set_ylabel('Número de Registros')
ax1.tick_params(axis='x', rotation=0)
# Agregar valores encima de las barras
for i, v in enumerate(season_counts.values):
ax1.text(i, v + 20, str(v), ha='center', va='bottom')
# Histograma de Weather (usando etiquetas)
weather_counts = bike_sharing_demand_df['weather_label'].value_counts()
weather_counts.plot(kind='bar', ax=ax2, color=['skyblue', 'gray', 'lightcoral', 'darkred'])
ax2.set_title('Distribución de Registros por Condición Climática')
ax2.set_xlabel('Condición Climática')
ax2.set_ylabel('Número de Registros')
ax2.tick_params(axis='x', rotation=0)
# Agregar valores encima de las barras
for i, v in enumerate(weather_counts.values):
ax2.text(i, v + 20, str(v), ha='center', va='bottom')
plt.tight_layout()
plt.show()
print("=== DISTRIBUCIÓN DE REGISTROS POR ESTACIÓN ===")
season_freq = bike_sharing_demand_df['season_label'].value_counts()
season_pct = bike_sharing_demand_df['season_label'].value_counts(normalize=True) * 100
for season in season_freq.index:
print(f"{season}: {season_freq[season]} registros ({season_pct[season]:.1f}%)")
print("\n=== DISTRIBUCIÓN DE REGISTROS POR CONDICIÓN CLIMÁTICA ===")
weather_freq = bike_sharing_demand_df['weather_label'].value_counts()
weather_pct = bike_sharing_demand_df['weather_label'].value_counts(normalize=True) * 100
for weather in weather_freq.index:
print(f"{weather}: {weather_freq[weather]} registros ({weather_pct[weather]:.1f}%)")=== DISTRIBUCIÓN DE REGISTROS POR ESTACIÓN ===
Winter: 2734 registros (25.1%)
Summer: 2733 registros (25.1%)
Autumn: 2733 registros (25.1%)
Spring: 2686 registros (24.7%)
=== DISTRIBUCIÓN DE REGISTROS POR CONDICIÓN CLIMÁTICA ===
Clear: 7192 registros (66.1%)
Fog: 2834 registros (26.0%)
Light Rain: 859 registros (7.9%)
Heavy Rain: 1 registros (0.0%)
# Tabla cruzada de frecuencias
cross_table = pd.crosstab(bike_sharing_demand_df['season_label'], bike_sharing_demand_df['weather_label'], margins=True)
print("\n=== TABLA CRUZADA: ESTACIÓN vs CONDICIÓN CLIMÁTICA ===")
print(cross_table)
# Heatmap de frecuencias (sin los totales)
plt.figure(figsize=(10, 6))
cross_table_no_margins = pd.crosstab(bike_sharing_demand_df['season_label'], bike_sharing_demand_df['weather_label'])
sns.heatmap(cross_table_no_margins, annot=True, cmap='Blues', fmt='d',
cbar_kws={'label': 'Número de Registros'})
plt.title('Distribución de Registros: Estación vs Condición Climática')
plt.xlabel('Condición Climática')
plt.ylabel('Estación del Año')
plt.show()=== TABLA CRUZADA: ESTACIÓN vs CONDICIÓN CLIMÁTICA ===
weather_label Clear Fog Heavy Rain Light Rain All
season_label
Autumn 1930 604 0 199 2733
Spring 1759 715 1 211 2686
Summer 1801 708 0 224 2733
Winter 1702 807 0 225 2734
All 7192 2834 1 859 10886
# Versión simple para replicar tu gráfico exacto
plt.figure(figsize=(12, 6))
hourly_avg = bike_sharing_demand_df.groupby('hour')['count'].mean()
plt.plot(hourly_avg.index, hourly_avg.values, 'o-', linewidth=2, markersize=6)
plt.grid(True, alpha=0.4)
plt.title('Rented bikes per hour of the day', fontsize=14, fontweight='bold')
plt.xlabel('hour of the day', fontsize=12)
plt.ylabel('Rented bikes (average)', fontsize=12)
plt.xticks(range(0, 24))
plt.show()
#how hourly bicycle use varies from season to season.
# Agrupar por hora y estación, y calcular la media
usage_by_season = bike_sharing_demand_df.groupby(['hour', 'season'], observed=True)['count'].mean().reset_index()
# Graficar
plt.figure(figsize=(10, 5))
sns.lineplot(data=usage_by_season, x='hour', y='count', hue='season')
plt.title('Promedio de bicicletas alquiladas por hora según estación')
plt.xlabel('Hora del día')
plt.ylabel('Bicicletas alquiladas (promedio)')
plt.xticks(range(0, 24))
plt.grid(True)
plt.legend(title='Season')
plt.show()
# Map season labels
season_mapping = {1: 'Spring', 2: 'Summer', 3: 'Fall', 4: 'Winter'}
bike_sharing_demand_df['season_name'] = bike_sharing_demand_df['season'].map(season_mapping)
# Group by day to get daily totals
daily_counts = bike_sharing_demand_df.groupby(['year', 'month', 'day', 'season_name'])['count'].sum().reset_index()
daily_counts.rename(columns={'count': 'daily_total'}, inplace=True)
# Individual histograms by season
fig, axes = plt.subplots(2, 2, figsize=(16, 12))
fig.suptitle('Daily Bike Sharing Records Distribution by Season',
fontsize=16, fontweight='bold', y=0.98)
axes = axes.flatten()
seasons = ['Spring', 'Summer', 'Fall', 'Winter']
colors = ['#7FB069', '#FFD23F', '#FF6B35', '#4ECDC4']
for i, season in enumerate(seasons):
data_season = daily_counts[daily_counts['season_name'] == season]['daily_total']
if len(data_season) > 0:
# Simple histogram
axes[i].hist(data_season, bins=20, color=colors[i], alpha=0.7,
edgecolor='black', linewidth=0.5)
# Title and labels
axes[i].set_title(f'{season}\n(σ = {data_season.std():.0f})',
fontweight='bold', fontsize=12)
axes[i].set_xlabel('Records per Day')
axes[i].set_ylabel('Frequency (days)')
axes[i].grid(True, alpha=0.3)
# Basic stats in corner
total_days = len(data_season)
total_bikes = data_season.sum()
mean_val = data_season.mean()
axes[i].text(0.02, 0.98,
f'Days: {total_days}\nTotal: {total_bikes:,} bikes\nAverage: {mean_val:.0f}/day',
transform=axes[i].transAxes, fontsize=9,
verticalalignment='top',
bbox=dict(boxstyle='round', facecolor='white', alpha=0.8))
plt.tight_layout()
plt.show()
# Combined comparison histogram
plt.figure(figsize=(14, 8))
for i, season in enumerate(seasons):
data_season = daily_counts[daily_counts['season_name'] == season]['daily_total']
if len(data_season) > 0:
plt.hist(data_season, bins=20, alpha=0.6,
label=f'{season} (n={len(data_season)})', color=colors[i])
plt.title('Daily Records Distribution Comparison by Season - Bike Sharing',
fontsize=14, fontweight='bold')
plt.xlabel('Records per Day')
plt.ylabel('Frequency (days)')
plt.legend()
plt.grid(True, alpha=0.3)
plt.show()
