Part I - Ford GoBike System Slideshow San Fransico Bay Area (February 2019)

by Oke Oladunsi

Table of Contents

• Brief Introduction
• reliminary Data Wrangling efforts
• Feature Engineering
• Selecting Features to work on
• Univariate Exploration
• Bivariate Exploration
• Multivariate Exploration
• Conclusions
• ReadMe Document
• Slideshow

Introduction

Bay Wheels (previously known as Ford GoBike) is a regional public bike sharing system in the San Francisco Bay Area, California. Bay Wheels is the first regional and large-scale bicycle sharing system deployed in California and on the West Coast of the United States with nearly 500,000 rides since the launch in 2017 and had about 10,000 annual subscribers as of January 2018.

The dataset used for this exploratory analysis consists of This dataset includes information about individual rides made in a bike-sharing system covering the Francisco Bay area of travels for the month February year 2019 for February 2019 alone in CSV format covering the greater San Francisco Bay area, also available here data for other cities.

Preliminary Wrangling

This data needs a little wrangling. so I would be changing some variables/features datatypes into appropriate ones using astype methed

# import all packages and set plots to be embedded inline
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sb
import datetime

%matplotlib inline

Gathering

ride_data = pd.read_csv("201902_fordgobike_tripdata.csv")
ride_data.head()

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	end_station_longitude	bike_id	user_type	member_birth_year	member_gender	bike_share_for_all_trip
0	52185	2019-02-28 17:32:10.1450	2019-03-01 08:01:55.9750	21.0	Montgomery St BART Station (Market St at 2nd St)	37.789625	-122.400811	13.0	Commercial St at Montgomery St	37.794231	-122.402923	4902	Customer	1984.0	Male	No
1	42521	2019-02-28 18:53:21.7890	2019-03-01 06:42:03.0560	23.0	The Embarcadero at Steuart St	37.791464	-122.391034	81.0	Berry St at 4th St	37.775880	-122.393170	2535	Customer	NaN	NaN	No
2	61854	2019-02-28 12:13:13.2180	2019-03-01 05:24:08.1460	86.0	Market St at Dolores St	37.769305	-122.426826	3.0	Powell St BART Station (Market St at 4th St)	37.786375	-122.404904	5905	Customer	1972.0	Male	No
3	36490	2019-02-28 17:54:26.0100	2019-03-01 04:02:36.8420	375.0	Grove St at Masonic Ave	37.774836	-122.446546	70.0	Central Ave at Fell St	37.773311	-122.444293	6638	Subscriber	1989.0	Other	No
4	1585	2019-02-28 23:54:18.5490	2019-03-01 00:20:44.0740	7.0	Frank H Ogawa Plaza	37.804562	-122.271738	222.0	10th Ave at E 15th St	37.792714	-122.248780	4898	Subscriber	1974.0	Male	Yes

Assessing the data

ride_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 183412 entries, 0 to 183411
Data columns (total 16 columns):
 #   Column                   Non-Null Count   Dtype  
---  ------                   --------------   -----  
 0   duration_sec             183412 non-null  int64  
 1   start_time               183412 non-null  object 
 2   end_time                 183412 non-null  object 
 3   start_station_id         183215 non-null  float64
 4   start_station_name       183215 non-null  object 
 5   start_station_latitude   183412 non-null  float64
 6   start_station_longitude  183412 non-null  float64
 7   end_station_id           183215 non-null  float64
 8   end_station_name         183215 non-null  object 
 9   end_station_latitude     183412 non-null  float64
 10  end_station_longitude    183412 non-null  float64
 11  bike_id                  183412 non-null  int64  
 12  user_type                183412 non-null  object 
 13  member_birth_year        175147 non-null  float64
 14  member_gender            175147 non-null  object 
 15  bike_share_for_all_trip  183412 non-null  object 
dtypes: float64(7), int64(2), object(7)
memory usage: 22.4+ MB

Cleaning

Checking if there are duplicated rows

ride_data.duplicated().sum()

0

Checking the amount of rows with empty values

# features with null values
null_columns = {}
for i in ride_data.columns.to_list():
    if ride_data[i].isnull().sum() > 0:
        null_columns[i] = ride_data[i].isnull().sum()
    else:
        pass
print(null_columns)

{'start_station_id': 197, 'start_station_name': 197, 'end_station_id': 197, 'end_station_name': 197, 'member_birth_year': 8265, 'member_gender': 8265}

Feature Engineering

creating new columns that would be needed for visualizations

Code

# Wrangling a little 
ride_data.dropna(inplace=True)
ride_data["member_birth_year"] = ride_data["member_birth_year"].astype("int")
ride_data["start_time"] = pd.to_datetime(ride_data["start_time"])
ride_data["end_time"] = pd.to_datetime(ride_data["end_time"])

More features are created

Code

# creating new colunms for time in year,month, day and hours
ride_data["year"] = ride_data["start_time"].dt.year
ride_data["month"] = ride_data["start_time"].dt.month
ride_data["days"] = ride_data["start_time"].dt.day
ride_data["boarding_hour"] = ride_data["start_time"].dt.hour

def getting_period_of_day(Hour):
    """ This function generates periods of the day from accepting numeric values in range 0-23 representing Morning-Afternoon-Evening-Night
    WHERE
    5am - 11am : Morning
    12am - 4pm : Afternoon
    5pm - 9pm : Evening
    10pm - 4am : Night
    """
    morning = [i for i in range(5,12)]
    afternoon = [i for i in range(12,17)]
    evening = [i for i in range(17,21)]
    night =  [i for i in range(21,24)] + [i for i in range(0,5)]

    if Hour in morning:
        return "morning"
    elif Hour in afternoon:
        return "afternoon"
    elif Hour in evening:
        return "evening"
    else:
        return "night"

def days(num):
    """ This function generates periods of the day from accepting numeric values in range 0-6 representing Sun- Sat
    WHERE
    0 : Sunday
    1 : Monday
    2 : Tuesday
    3 : Wednesday
    4 : Thursday
    5 : Friday
    6 : Saturday

    """
    days = {0:"Sunday",1:"Monday",2:"Tuesday",3:"Wednesday",4:"Thursday",5:"Friday",6:"Saturday"}
    return days.get(num)

ride_data["period_of_day"] = ride_data["boarding_hour"].apply(lambda x: getting_period_of_day(x))
ride_data["day_of_week"] = ride_data["start_time"].dt.weekday.apply(lambda x: days(x))

NB: The essence of the feature engineering done above to enable be provide insights into how to be effectively maximise both human and non-human resources to increase profit and at the same time provide premium bike sharing service

Test

ride_data["year"].value_counts()

2019    174952
Name: year, dtype: int64

• this is an affirmation that we only have data of the year 2019
• the code block below shows that it only for the month of ` FEBRUARY`

ride_data["month"].value_counts()

2    174952
Name: month, dtype: int64

# pragrammatically assessing the data
ride_data["period_of_day"].value_counts()

morning      65252
evening      52657
afternoon    46114
night        10929
Name: period_of_day, dtype: int64

• from the above cell we can see that in all our stations we have more rides in morning
• the night has the least travel

ride_data["day_of_week"].value_counts()

Wednesday    33712
Monday       30584
Tuesday      28426
Thursday     27663
Sunday       25641
Saturday     14512
Friday       14414
Name: day_of_week, dtype: int64

• the cell above shows that people travel most on wednesday
• Friday has the least amount of travel

# confirming if there is the stations in the start_station_name are the same as the ones in the end_station_name 
sorted(ride_data["start_station_name"].value_counts().index) == sorted(ride_data["end_station_name"].value_counts().index)

True

• confirming if the same stations in the start stations(onboarding) are all present in the end stations(alighting)

ride_data["member_gender"].value_counts()

Male      130500
Female     40805
Other       3647
Name: member_gender, dtype: int64

• the amount of rides taken by men are 3x that of women and 35x that of Other genders

ride_data["user_type"].value_counts()

Subscriber    158386
Customer       16566
Name: user_type, dtype: int64

• occassionally we have riders that aren't subcribers making used of the service but subcribers are significantly more that occassional riders
  we can probe how long occasional riders spend using the service compared to subscribers

ride_data["bike_share_for_all_trip"].value_counts()

No     157606
Yes     17346
Name: bike_share_for_all_trip, dtype: int64

• this shows that most riders don't like the idea of sharing the rides with other users

Code

ride_data["age"] = 2019 - ride_data["member_birth_year"]

Code

ride_data["user_type"] = ride_data["user_type"].astype("category")
ride_data["member_gender"] = ride_data["member_gender"].astype("category")
ride_data["boarding_hour"] = ride_data["boarding_hour"].astype("category")
ride_data["period_of_day"] = ride_data["period_of_day"].astype("category")
ride_data["day_of_week"] = ride_data["day_of_week"].astype("category")

ride_data["start_station_id"] =ride_data["start_station_id"].astype('str')
ride_data["end_station_id"] = ride_data["end_station_id"].astype('str')
ride_data["bike_id"] =ride_data["bike_id"].astype('str')

Test

ride_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 174952 entries, 0 to 183411
Data columns (total 23 columns):
 #   Column                   Non-Null Count   Dtype         
---  ------                   --------------   -----         
 0   duration_sec             174952 non-null  int64         
 1   start_time               174952 non-null  datetime64[ns]
 2   end_time                 174952 non-null  datetime64[ns]
 3   start_station_id         174952 non-null  object        
 4   start_station_name       174952 non-null  object        
 5   start_station_latitude   174952 non-null  float64       
 6   start_station_longitude  174952 non-null  float64       
 7   end_station_id           174952 non-null  object        
 8   end_station_name         174952 non-null  object        
 9   end_station_latitude     174952 non-null  float64       
 10  end_station_longitude    174952 non-null  float64       
 11  bike_id                  174952 non-null  object        
 12  user_type                174952 non-null  category      
 13  member_birth_year        174952 non-null  int64         
 14  member_gender            174952 non-null  category      
 15  bike_share_for_all_trip  174952 non-null  object        
 16  year                     174952 non-null  int64         
 17  month                    174952 non-null  int64         
 18  days                     174952 non-null  int64         
 19  boarding_hour            174952 non-null  category      
 20  period_of_day            174952 non-null  category      
 21  day_of_week              174952 non-null  category      
 22  age                      174952 non-null  int64         
dtypes: category(5), datetime64[ns](2), float64(4), int64(6), object(6)
memory usage: 26.2+ MB

What is the structure of your dataset?

This dataset initially consists of a total of 183412 rows and 16 columns where six of th 16 columns are having null values you can check the cell above for the amount of null entries for each feature. Then after I had wrangled the data it has reduced to 174952 rows and the columns increased to 21 columns. This dataset is for rides in the month of february 2019

What is/are the main feature(s) of interest in your dataset?

I'm most interested in exploring the bike trips' duration and rental events occurrance patterns, along with how these relate to the riders' characteristics, i.e. their user type, gender, age, etc, to get a sense of how and what people are using the bike sharing service for. Sample questions to answer: When are most trips taken in terms of time of day, day of the week, or month of the year? How long does the average trip take? Does the above depend on if a user is a subscriber or customer?

What features in the dataset do you think will help support your investigation into your feature(s) of interest?

member_birth_year,start_station_latitude,start_station_longitude,end_station_latitude,start_station_latitude are features i think i will be needing for this visualizations though I might probably create new ones to support these one I currently have.

numeric_colunms = ["duration_sec","start_station_latitude","start_station_longitude","end_station_latitude","end_station_longitude","age",]
categorical_columns = ["user_type","member_gender","period_of_day","day_of_week"]

Univariate Exploration

ride_data.describe()

	duration_sec	start_station_latitude	start_station_longitude	end_station_latitude	end_station_longitude	member_birth_year	year	month	days	age
count	174952.000000	174952.000000	174952.000000	174952.000000	174952.000000	174952.000000	174952.0	174952.0	174952.000000	174952.000000
mean	704.002744	37.771220	-122.351760	37.771414	-122.351335	1984.803135	2019.0	2.0	15.312337	34.196865
std	1642.204905	0.100391	0.117732	0.100295	0.117294	10.118731	0.0	0.0	8.033926	10.118731
min	61.000000	37.317298	-122.453704	37.317298	-122.453704	1878.000000	2019.0	2.0	1.000000	18.000000
25%	323.000000	37.770407	-122.411901	37.770407	-122.411647	1980.000000	2019.0	2.0	8.000000	27.000000
50%	510.000000	37.780760	-122.398279	37.781010	-122.397437	1987.000000	2019.0	2.0	15.000000	32.000000
75%	789.000000	37.797320	-122.283093	37.797673	-122.286533	1992.000000	2019.0	2.0	22.000000	39.000000
max	84548.000000	37.880222	-121.874119	37.880222	-121.874119	2001.000000	2019.0	2.0	28.000000	141.000000

Checking how correlated each numeric columns are to each other

def show_numeric_datapoints_corr():
        plt.figure(figsize = [8, 5])
        sb.heatmap(ride_data[numeric_colunms].corr(), annot = True, fmt = '.3f',
                cmap = 'vlag_r', center = 0)
        plt.title("How each numeric variables relates\n")
        plt.show()
        

show_numeric_datapoints_corr()

• just exploring visually how correlated each numeric columns are to each other

np.arange(100,2000+100,100)//60

array([ 1,  3,  5,  6,  8, 10, 11, 13, 15, 16, 18, 20, 21, 23, 25, 26, 28,
       30, 31, 33])

ride_data["travelTime_minutes"] = ride_data["duration_sec"]//60
ride_data["travelTime_hours"] = ride_data["duration_sec"]//(60**2)

What is the distribution of travel time and ages of riders

• are they normally distribute or skewed?

def numeric_plots():
    plt.figure(figsize=[20,5])

    plt.subplot(1,3,1)
    bins = np.arange(100,2000+100,100)
    sb.histplot(data=ride_data,x ="duration_sec",bins=bins);
    plt.xticks(bins,rotation=45)
    plt.title("Distributon of time travel in (secs)")  

    plt.subplot(1,3,2)
    min_bins = np.arange(0,60,5)
    sb.histplot(data=ride_data,x ="travelTime_minutes",bins=min_bins);
    plt.xticks(min_bins,rotation=45)
    plt.title("Distributon of time travel in (mins)")

    plt.subplot(1,3,3)
    hour_bins = np.arange(0,ride_data["travelTime_hours"].max(),1)
    sb.histplot(data=ride_data,x ="travelTime_hours",bins=hour_bins);
    plt.xticks(hour_bins,rotation=45)
    plt.title("Distributon of time travel in (hours)")

    plt.figure(figsize=[8,5])
    bins = np.arange(15,80+5,5)
    sb.histplot(data=ride_data,x ="age",bins=bins);
    plt.xticks(bins,rotation=45)
    plt.title("Distributon of passeger ages")


numeric_plots()

ride_data["travelTime_hours"].value_counts()

0     173566
1        848
2        208
3         91
4         46
5         35
9         17
6         17
7         13
14        13
17        13
8         12
13        11
10        10
12         9
19         7
15         7
20         6
11         6
18         5
16         5
23         4
22         3
Name: travelTime_hours, dtype: int64

new= ride_data["travelTime_minutes"].value_counts().to_frame()[:20]
new["percent"] = ride_data["travelTime_minutes"].value_counts()[:20]/ride_data["travelTime_minutes"].value_counts()[:30].sum()

ride_data[ride_data["travelTime_minutes"] < 2]

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours
43	116	2019-02-28 23:44:00.988	2019-02-28 23:45:57.482	104.0	4th St at 16th St	37.767045	-122.390833	93.0	4th St at Mission Bay Blvd S	37.770407	...	No	2019	2	28	23	night	Wednesday	60	1	0
72	113	2019-02-28 23:26:28.335	2019-02-28 23:28:21.625	163.0	Lake Merritt BART Station	37.797320	-122.265320	163.0	Lake Merritt BART Station	37.797320	...	No	2019	2	28	23	night	Wednesday	44	1	0
85	73	2019-02-28 23:17:06.483	2019-02-28 23:18:19.711	67.0	San Francisco Caltrain Station 2 (Townsend St...	37.776639	-122.395526	80.0	Townsend St at 5th St	37.775235	...	No	2019	2	28	23	night	Wednesday	24	1	0
128	71	2019-02-28 22:59:32.741	2019-02-28 23:00:44.152	280.0	San Fernando St at 7th St	37.337122	-121.883215	310.0	San Fernando St at 4th St	37.335885	...	Yes	2019	2	28	22	night	Wednesday	24	1	0
251	83	2019-02-28 22:16:36.506	2019-02-28 22:17:59.632	263.0	Channing Way at San Pablo Ave	37.862827	-122.290231	263.0	Channing Way at San Pablo Ave	37.862827	...	No	2019	2	28	22	night	Wednesday	24	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
183252	94	2019-02-01 06:14:43.739	2019-02-01 06:16:18.130	350.0	8th St at Brannan St	37.771431	-122.405787	79.0	7th St at Brannan St	37.773492	...	No	2019	2	1	6	morning	Thursday	24	1	0
183305	100	2019-02-01 04:46:54.805	2019-02-01 04:48:34.843	80.0	Townsend St at 5th St	37.775235	-122.397437	67.0	San Francisco Caltrain Station 2 (Townsend St...	37.776639	...	No	2019	2	1	4	night	Thursday	69	1	0
183345	98	2019-02-01 02:09:43.393	2019-02-01 02:11:22.174	133.0	Valencia St at 22nd St	37.755213	-122.420975	127.0	Valencia St at 21st St	37.756708	...	No	2019	2	1	2	night	Thursday	20	1	0
183395	95	2019-02-01 00:37:23.115	2019-02-01 00:38:58.346	276.0	Julian St at The Alameda	37.332233	-121.912516	277.0	Morrison Ave at Julian St	37.333658	...	Yes	2019	2	1	0	night	Thursday	26	1	0
183405	111	2019-02-01 00:14:49.874	2019-02-01 00:16:41.301	324.0	Union Square (Powell St at Post St)	37.788300	-122.408531	19.0	Post St at Kearny St	37.788975	...	No	2019	2	1	0	night	Thursday	35	1	0

2979 rows × 25 columns

new["travelTime_minutes"].sum()/ride_data.shape[0]

0.9187777218894325

ride_data[ride_data["travelTime_hours"] == ride_data["travelTime_hours"].max()]

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours
5203	83195	2019-02-27 14:47:23.181	2019-02-28 13:53:58.433	243.0	Bancroft Way at College Ave	37.869360	-122.254337	248.0	Telegraph Ave at Ashby Ave	37.855956	...	Yes	2019	2	27	14	afternoon	Tuesday	57	1386	23
85465	84548	2019-02-16 15:48:25.029	2019-02-17 15:17:33.080	3.0	Powell St BART Station (Market St at 4th St)	37.786375	-122.404904	368.0	Myrtle St at Polk St	37.785434	...	No	2019	2	16	15	afternoon	Friday	38	1409	23
112435	83407	2019-02-11 16:25:33.069	2019-02-12 15:35:40.956	77.0	11th St at Natoma St	37.773507	-122.416040	344.0	16th St Depot	37.766349	...	No	2019	2	11	16	afternoon	Sunday	31	1390	23
127999	83519	2019-02-09 15:16:17.537	2019-02-10 14:28:17.270	72.0	Page St at Scott St	37.772406	-122.435650	43.0	San Francisco Public Library (Grove St at Hyde...	37.778768	...	No	2019	2	9	15	afternoon	Friday	29	1391	23

4 rows × 25 columns

• the plot above shows the distribution of duration in sec and we can see that the distrubtion is rigth skewed and most people spend between travel 200-600ses to get to their various destinations

• Usually most rides take about 20mins ride from a station to another and while the travel time in hours shows that usually the time during rides are
  within one hour from boarding station to alighting station

• the second plot shows that most people that uses the ride services are within the ages of 25-40. while the maximum age of 141 we saw in our describe probably is an outlier.

• after careful examination of the duration sec column we could see that some ride last for 23 hour which is actually strange which will make me try to visualize duration of travel in hours. ## NB: What could cause the large travel time in the dataset. Therefore from now on I wiill based my visuaizations on rides withing one hour

ride_data = ride_data[ride_data["travelTime_hours"] <= 60]

def rider_share():
    plt.figure(figsize=[10,8])
    base_color = sb.color_palette()[0]
    sb.countplot(data=ride_data,x="user_type",color=base_color)
    plt.title("How does occassional rider make use of the ride service")

    print(ride_data.user_type.value_counts())
rider_share()

Subscriber    158386
Customer       16566
Name: user_type, dtype: int64

• Above we see that we have more subscribers of about (10x more) using the ride service than occassional custormers

def gender():
    plt.figure(figsize=[10,8])
    base_color = sb.color_palette()[0]
    sb.countplot(data=ride_data,x="member_gender",color = base_color)
    plt.title("Which gender make use of the service the most")
gender()

• the dataset shows that we have more males making use of the service than other genders

def observing_Week_days():
    plt.figure(figsize=[8,5])
    base_color = sb.color_palette()[0]
    weekday = [ 'Sunday', 'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday']
    sb.countplot(data=ride_data,x="day_of_week",color=base_color,order=weekday)
    plt.xticks(rotation=45)
    plt.title("Which Day of the week does the service experience the most surge/request")

observing_Week_days()

• The dataset shows that less people use the bike share service during the weekend espectially on staurdays and sundays while wednesday is when people travels the most

def get_share():
    plt.figure(figsize=[8,5])
    base_color = sb.color_palette()[0]
    sb.countplot(data=ride_data,x="bike_share_for_all_trip",color=base_color)
    plt.title("how does riders react to ride sharing feature")

get_share()

• I was expecting more riders not wanting not sharing the rides because from the previous chart we had more men using the platform to travel and men have the tendecy to need privacy more than females and other genders

Discuss the distribution(s) of your variable(s) of interest. Were there any unusual points? Did you need to perform any transformations?

The duration_sec is not normally distributed but skewed. Then I create new feature travel_minutes to enable me observe it more closely which shows that the majority of the members did not use bike share for all of their trips, and most were around 25 to 40 years old. Most rides were quick and short, lasted between 5 to 10 minutes, that some riders travel time is close to 23hrs which is not normal.

Of the features you investigated, were there any unusual distributions? Did you perform any operations on the data to tidy, adjust, or change the form of the data? If so, why did you do this?

Yes, travelTime_hour shows that 99% of rides are within 1hr and about 91% of these rides are within 20mins. there are some features that have unusual distributions like the ages of rides that holds a values of 141 years that seems odd.

Bivariate Exploration

In this section, investigate relationships between pairs of variables in your data. Make sure the variables that you cover here have been introduced in some fashion in the previous section (univariate exploration).

ride_data[ride_data["travelTime_minutes"] > 660]["day_of_week"].value_counts()

Sunday       15
Tuesday      15
Friday       14
Saturday     12
Monday       11
Thursday     11
Wednesday    11
Name: day_of_week, dtype: int64

one__hour = ride_data[ride_data["travelTime_minutes"] < 60]
def categorical__timeTravel(x, y, **kwargs):
    """ Quick hack for creating box plots with seaborn's PairGrid. """
    default_color = sb.color_palette()[0]
    
    sb.boxplot(x, y, color = default_color)
    plt.title("{} v {}".format(y.name,x.name))



g = sb.PairGrid(data = one__hour, y_vars =[ 'age','travelTime_minutes'], x_vars = ["user_type","member_gender","period_of_day"],
                size = 3, aspect = 1.5)
g.map(categorical__timeTravel)
plt.show();

/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/axisgrid.py:1152: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(UserWarning(msg))
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning

def more_categorical__timeTravel(x,y, **kwargs):
    """ Quick hack for creating violin plots with seaborn's PairGrid. """
    default_color = sb.color_palette()[0]
    sb.violinplot(x, y, color = default_color)
    plt.title("{} v {}".format(y.name,x.name))


g = sb.PairGrid(data = one__hour, y_vars ="travelTime_minutes", x_vars = ["user_type","member_gender","period_of_day"],
                size = 3, aspect = 1.5)
plt.figure(figsize = [20, 15])
g.map(more_categorical__timeTravel)
plt.show();

/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/axisgrid.py:1152: UserWarning: The `size` parameter has been renamed to `height`; please update your code.
  warnings.warn(UserWarning(msg))
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning
/home/oldunsi/anaconda3/envs/Iqube/lib/python3.7/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  FutureWarning

<Figure size 1440x1080 with 0 Axes>

NB This chart above are for rides within One hour (1h)

• we can see that there are no significant difference in the avagrage age of customer and subscribers
• yet there is a slightly higher age difference in other genders compared to both male and female
• finally on ages, we can see that younger people travel more at night.

#

• we can see that there is significant difference in the avagrage time travel by customer(13mins) and subscribers(8mins). where is seems subscribers tend to know where they going hereby reducing travel time.
• men tend to travel on the average than other genders
• finally people that travel at night which are most young people have a specific place they are going to by the fact that average rides spend less than 10mins to their destinations.

def riders__weekdays():
    base_color = sb.color_palette()[0]
    weekday = [ 'Sunday', 'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday']
    sb.barplot(data=one__hour, x='day_of_week', y='travelTime_minutes',ci='sd', color=base_color,order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute per day');
    plt.xticks(rotation=45)
    plt.title("how does riders react to ride sharing feature")

riders__weekdays()

• the chart above shows that people travel time from monday through thursday are within 10 mins but on friday and saturday the avg time goes above 10 mins which occassionaly goes above 15 min but not more than 17mins. Yet during the weekends we have some rides that goes beyond 17mins int 20mins

ride_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 174952 entries, 0 to 183411
Data columns (total 25 columns):
 #   Column                   Non-Null Count   Dtype         
---  ------                   --------------   -----         
 0   duration_sec             174952 non-null  int64         
 1   start_time               174952 non-null  datetime64[ns]
 2   end_time                 174952 non-null  datetime64[ns]
 3   start_station_id         174952 non-null  object        
 4   start_station_name       174952 non-null  object        
 5   start_station_latitude   174952 non-null  float64       
 6   start_station_longitude  174952 non-null  float64       
 7   end_station_id           174952 non-null  object        
 8   end_station_name         174952 non-null  object        
 9   end_station_latitude     174952 non-null  float64       
 10  end_station_longitude    174952 non-null  float64       
 11  bike_id                  174952 non-null  object        
 12  user_type                174952 non-null  category      
 13  member_birth_year        174952 non-null  int64         
 14  member_gender            174952 non-null  category      
 15  bike_share_for_all_trip  174952 non-null  object        
 16  year                     174952 non-null  int64         
 17  month                    174952 non-null  int64         
 18  days                     174952 non-null  int64         
 19  boarding_hour            174952 non-null  category      
 20  period_of_day            174952 non-null  category      
 21  day_of_week              174952 non-null  category      
 22  age                      174952 non-null  int64         
 23  travelTime_minutes       174952 non-null  int64         
 24  travelTime_hours         174952 non-null  int64         
dtypes: category(5), datetime64[ns](2), float64(4), int64(8), object(6)
memory usage: 32.9+ MB

def more_insight_on_riders_behavour():
    plt.figure(figsize = [20, 5])
    period = ["morning","afternoon","evening","night"]
    plt.subplot(1,2,1)
    sb.countplot(data=ride_data, x='period_of_day',palette="Blues",hue="user_type", order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per user type based on period of day")

    plt.subplot(1,2,2)
    sb.countplot(data=ride_data, x='period_of_day',palette="Blues",hue="member_gender",order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per gender based period of day")

    plt.figure(figsize = [20, 8])
    plt.subplot(1,2,1)
    sb.countplot(data=ride_data, x='boarding_hour',hue="user_type",palette="Greens");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per user type on hourly basis")

    plt.subplot(1,2,2)
    sb.countplot(data=ride_data, x='boarding_hour',hue="member_gender",palette="Greens");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per gender on hourly basis")


    weekday = [ 'Sunday', 'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday']
    

    plt.figure(figsize = [20, 8])
    plt.subplot(1,2,1)
    sb.countplot(data=ride_data, x='day_of_week',hue="user_type",palette="Blues",order=weekday);
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per user type based on weekday")

    plt.subplot(1,2,2)
    sb.countplot(data=ride_data, x='day_of_week',hue="member_gender",palette="Blues",order=weekday);
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("entire dataset riders riding pattern per gender based weekday")


    

more_insight_on_riders_behavour()

• The first row of chart show that on any day of week we tend to have more subscriber using the service in the morning than at any period of the day. the visuals also shows that the distribution varies from subscriber to occassional customers where customers tend to use the service less in the evening than in the afternoon but subscribers use the service more in the evening than in the afternoon. on the second chart of the first row we can see that there seem to identical distribution for male and female but other genders seems to have constant trend through morning till evening but reduced at night.
• The busiest hours are pretty much consistent throughout the month with rush hours occuring around 8AM and 5PM.
• Wednesday is the day the ride service is used the most and during the weekend there is a constant reduction of usage.

def rides_in_feb():
    plt.figure(figsize = [20, 8])
    plt.subplot(1,2,1)
    sb.countplot(data=ride_data, x='days',hue="user_type",palette="Blues");
    plt.xlabel('total per day in february 2019');
    plt.ylabel('Count');
    plt.yticks(np.arange(0,9000,1000),np.arange(0,9000,1000))
    plt.title("entire dataset riders riding pattern per user type through Feb 2019")

    plt.subplot(1,2,2)
    sb.countplot(data=ride_data, x='days',hue="member_gender",palette="Blues");
    plt.xlabel('total per day in february 2019');
    plt.ylabel('Count');
    plt.yticks(np.arange(0,9000,1000),np.arange(0,9000,1000))
    plt.title("entire dataset riders riding pattern per gender through Feb 2019 ")

rides_in_feb()

• throught the month customers seems to be more constitent in the usage the service than subscribers
• the distribution of rides shows that female riders have a normally distributed shape than other genders

Talk about some of the relationships you observed in this part of the investigation. How did the feature(s) of interest vary with other features in the dataset?

There are a lot more subscriber usage than customers. The riding habit/pattern varies a lot between subscribers and customers.
Subscribers use the bike sharing system for work commnute thus most trips were on work days (Mon-Fri) and especially during rush hours
(when going to work in the morning (8am) and getting off work in the evening (5pm)).

Did you observe any interesting relationships between the other features (not the main feature(s) of interest)?

There is significant difference in the avagrage time travel by customer(13mins) and subscribers(8mins). where is seems subscribers tend to know where they going hereby reducing travel time.

People travel time from monday through thursday are within 10 mins but on friday and saturday the avg time goes above 10 mins which occassionaly goes above 15 min but not more than 17mins. Yet during the weekends we have some rides that goes beyond 17mins int 20mins

Multivariate Exploration

Create plots of three or more variables to investigate your data even further. Make sure that your investigations are justified, and follow from your work in the previous sections.

weekday = [ 'Sunday', 'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday']
period = ["morning","afternoon","evening","night"]

def multivariate():
    
    plt.figure(figsize = [20,6])
    plt.subplot(1,3,1)
    sb.pointplot(data=ride_data, x='day_of_week', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("riders riding pattern per user type on weekday basis")

    plt.subplot(1,3,2)
    sb.pointplot(data=ride_data, x='boarding_hour', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("riders riding pattern per user type on hourly basis")


    plt.subplot(1,3,3)
    sb.pointplot(data=ride_data, x='days', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Riders riding pattern throughout Feb 2019 per user type")



    
    plt.figure(figsize = [20, 8])
    plt.subplot(2,3,1)
    sb.pointplot(data=ride_data, x='day_of_week', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("riders riding pattern per gender on weekday basis")

    plt.subplot(2,3,2)
    sb.pointplot(data=ride_data, x='boarding_hour', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Hours');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("riders riding pattern per gender on hourly basis")


    plt.subplot(2,3,3)
    sb.pointplot(data=ride_data, x='days', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Riders riding pattern throughout Feb 2019 per gender")

    
    

    
#     
multivariate()

• We see that both subcribers and customers riders spend more time riding from thursday through saturday but subcribers are consistent with the 10mins travel from Sun-Wed,whereas customers tend to ride for fun in the afternoon or early evenings over weekends.

• like I was expecting, people that don't have a subcription plan tend to spend more time riding.

• That dataset also shows that even though all genders incresed their travel time during weekend male riders travel time reduces by saturday

• we can see that other genders spend more time riding towards the end of the month

We have been looking at the entire dataset since, But What about checking out how Younger riders riding patterns differs from older riders

• The charts below should be able to give insights into the each categories of people.

  • Young riders are within the ages of 18yrs-40yrs
    
  • Older riders are are within the ages of 60yrs and above

ride_youth = ride_data[ride_data["age"].isin(np.arange(18,40,1))]

def youthriders():
    plt.figure(figsize = [20,5])
    plt.subplot(1,3,1)
    sb.pointplot(data=ride_youth, x='day_of_week', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Young riders riding pattern per user type on weekday basis")

    plt.subplot(1,3,2)
    sb.pointplot(data=ride_youth, x='boarding_hour', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Hours');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Young riders riding pattern per user type on hourly basis")


    plt.subplot(1,3,3)
    sb.pointplot(data=ride_youth, x='days', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("How Young users make use of the service throughout Feb 2019 \n per user type")

    plt.figure(figsize = [20,10])
    plt.subplot(2,3,1)
    sb.pointplot(data=ride_youth, x='day_of_week', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Young riders riding pattern per gender on weekly basis")

    plt.subplot(2,3,2)
    sb.pointplot(data=ride_youth, x='boarding_hour', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Hours');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Young riders riding pattern per gender on hourly basis")


    plt.subplot(2,3,3)
    sb.pointplot(data=ride_youth, x='days', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Young riders rinding patterns throughout Feb 2019 per gender")

    
youthriders()

NB: The chart of youth riding pattern and that of the general looks the same.

• user within the ages of 18-40 spend about 10mins on average subscribers in 24hrs
• on the weekly chart we can see that the youth trend is similar to that of the whole data where both subcribers and customers spend more travel time riding during the weekends
• throughout the month of february young subcribers spend about 10mins on riding to their destinations but customers are travel time is not constant.

weekend_ride = ride_data[ride_data["day_of_week"].isin(['Friday', 'Saturday'])]
prop = weekend_ride["member_gender"].value_counts()/weekend_ride["member_gender"].value_counts().sum()

def weekendgender():

    tick_props = np.arange(0,prop.max(),0.1)
    tick_names =["{:0.2f}".format(v) for v  in tick_props]
    base_color = sb.color_palette()[0]

    plt.figure(figsize = [20,20])
    plt.subplot(3,2,1)
    sb.countplot(data=weekend_ride,x="member_gender",color = base_color)
    plt.title("Which gender make use of the service the most during weekends")
    plt.yticks(tick_props*weekend_ride["member_gender"].value_counts().sum(),tick_names)
    
    plt.subplot(3,2,2)
    sb.countplot(data=weekend_ride,x="boarding_hour",color=base_color)
    plt.xticks(rotation=45)
    plt.title("What time Day of does the service experience the most surge/request")


    ###################################

    plt.figure(figsize = [20,10])
    plt.subplot(2,2,1)   
    sb.countplot(data=weekend_ride, x='boarding_hour',hue="member_gender",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("What time does each gender make use of the service the most during weekends")
    
    plt.subplot(2,2,2)
    sb.countplot(data=weekend_ride, x='boarding_hour',hue="user_type",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("Is there a change in usage of service during weekends: focus (User_type)")

    ################################
    

weekendgender()

def more_visual_weekend():
    plt.figure(figsize = [20, 8])
    plt.subplot(1,2,1)
    sb.pointplot(data=weekend_ride, x='boarding_hour', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Hours of Day');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("What time of day does each gender make use of the service the most during weekends") 

    

    plt.subplot(1,2,2)
    sb.pointplot(data=weekend_ride, x='boarding_hour', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Hours of Day');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("What time of day does each user type make use of the service the most during weekends") 
more_visual_weekend()

weekend_ride.head()

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours
29922	70925	2019-02-24 07:08:31.270	2019-02-25 02:50:36.590	375.0	Grove St at Masonic Ave	37.774836	-122.446546	71.0	Broderick St at Oak St	37.773063	...	No	2019	2	24	7	morning	Saturday	30	1182	19
29947	3788	2019-02-24 23:59:28.080	2019-02-25 01:02:36.220	99.0	Folsom St at 15th St	37.767037	-122.415442	386.0	24th St at Bartlett St	37.752105	...	Yes	2019	2	24	23	night	Saturday	38	63	1
29962	1080	2019-02-24 23:57:56.631	2019-02-25 00:15:57.338	189.0	Genoa St at 55th St	37.839649	-122.271756	339.0	Jackson St at 11th St	37.800002	...	No	2019	2	24	23	night	Saturday	29	18	0
29966	1131	2019-02-24 23:52:51.701	2019-02-25 00:11:43.648	133.0	Valencia St at 22nd St	37.755213	-122.420975	109.0	17th St at Valencia St	37.763316	...	No	2019	2	24	23	night	Saturday	39	18	0
29969	1667	2019-02-24 23:42:23.007	2019-02-25 00:10:10.068	249.0	Russell St at College Ave	37.858473	-122.253253	249.0	Russell St at College Ave	37.858473	...	No	2019	2	24	23	night	Saturday	21	27	0

5 rows × 25 columns

youth_weekend_ride = weekend_ride[weekend_ride["age"].isin(np.arange(18,40,1))]

def youthweekendgender():
    
    plt.figure(figsize=[20,8])
    plt.subplot(1,2,1)
    sb.countplot(data=youth_weekend_ride, x='period_of_day',palette="Blues",hue="user_type", order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("Period of day does each User type make use of the service the most during weekends: \n age (18-39)") 

    plt.subplot(1,2,2)
    sb.countplot(data=youth_weekend_ride, x='period_of_day',palette="Blues",hue="member_gender",order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("What Period of day does each gender make use of the service the most during weekends:\n age (18-39)") 


    plt.figure(figsize = [20, 8])
    plt.subplot(2,2,1)
    sb.countplot(data=youth_weekend_ride, x='boarding_hour',hue="user_type",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("Is there a change in usage of service during weekends: focus (User_type) age (18-39)")

    plt.subplot(2,2,2)
    sb.countplot(data=youth_weekend_ride, x='boarding_hour',hue="member_gender",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("time each gender make use of the service the most during weekends: age (18-39)")   


youthweekendgender()

def more_Youth_weekend_rides():
    plt.figure(figsize = [20, 8])
    plt.subplot(1,2,1)
    sb.pointplot(data=youth_weekend_ride, x='boarding_hour', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Hours of Day');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title(" Avg. time travel in (mins) for each gender on hourly basis during weekend: age (18-39)");
    

    plt.subplot(1,2,2)
    sb.pointplot(data=youth_weekend_ride, x='boarding_hour', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Hours of Day');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("Avg. time travel in (mins) for user type based on hourly basis during weekend: age (18-39)");

more_Youth_weekend_rides()

• The user type feature looks skewed to the right showing users use the service more in the morning and afternoon than at night
• but the hourly usage is normally distributed. We can see that there seem to a normal distributed fig from 6am to 11pm
• Usage of the service peaked at 2pm using user_types i.e (whether a user is subscriber or not)
• Using gender the peak period is also 2pm in the afternoon

NB: hypothesis ; Older riders should experience a reverse of the trend we have been noticing

ride_old = ride_data[ride_data["age"] >= 60]

weekend_ride_old = ride_old[ride_old["day_of_week"].isin(['Friday', 'Saturday'])]
weekend_ride_old

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours
30089	397	2019-02-24 22:12:17.173	2019-02-24 22:18:54.980	127.0	Valencia St at 21st St	37.756708	-122.421025	76.0	McCoppin St at Valencia St	37.771662	...	Yes	2019	2	24	22	night	Saturday	74	6	0
30329	198	2019-02-24 20:37:28.392	2019-02-24 20:40:46.840	175.0	49th St at Telegraph Ave	37.835946	-122.262366	210.0	45th St at Manila	37.833294	...	No	2019	2	24	20	evening	Saturday	68	3	0
30356	578	2019-02-24 20:22:18.170	2019-02-24 20:31:56.230	251.0	California St at University Ave	37.870555	-122.279720	251.0	California St at University Ave	37.870555	...	Yes	2019	2	24	20	evening	Saturday	63	9	0
30405	487	2019-02-24 20:06:13.706	2019-02-24 20:14:21.570	42.0	San Francisco City Hall (Polk St at Grove St)	37.778650	-122.418230	60.0	8th St at Ringold St	37.774520	...	No	2019	2	24	20	evening	Saturday	74	8	0
30453	673	2019-02-24 19:49:22.307	2019-02-24 20:00:35.751	24.0	Spear St at Folsom St	37.789677	-122.390428	79.0	7th St at Brannan St	37.773492	...	Yes	2019	2	24	19	evening	Saturday	76	11	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
177035	779	2019-02-02 09:01:20.129	2019-02-02 09:14:19.222	230.0	14th St at Mandela Pkwy	37.810743	-122.291415	315.0	Market St at 45th St	37.834174	...	No	2019	2	2	9	morning	Friday	62	12	0
177114	324	2019-02-02 08:11:11.241	2019-02-02 08:16:35.327	186.0	Lakeside Dr at 14th St	37.801319	-122.262642	200.0	2nd Ave at E 18th St	37.800214	...	No	2019	2	2	8	morning	Friday	60	5	0
177150	1685	2019-02-02 06:58:46.187	2019-02-02 07:26:51.784	61.0	Howard St at 8th St	37.776513	-122.411306	116.0	Mississippi St at 17th St	37.764802	...	No	2019	2	2	6	morning	Friday	62	28	0
177155	2820	2019-02-02 06:19:05.511	2019-02-02 07:06:06.446	131.0	22nd St at Dolores St	37.755000	-122.425728	41.0	Golden Gate Ave at Polk St	37.781270	...	No	2019	2	2	6	morning	Friday	61	47	0
177160	1124	2019-02-02 06:23:23.725	2019-02-02 06:42:08.261	80.0	Townsend St at 5th St	37.775235	-122.397437	66.0	3rd St at Townsend St	37.778742	...	No	2019	2	2	6	morning	Friday	69	18	0

769 rows × 25 columns

def oldriders():
       
    plt.figure(figsize = [20, 5])
    plt.subplot(1,2,1)
    sb.pointplot(data=ride_old, x='day_of_week', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("weekday older people riding pattern: age > 60yrs")
    
    plt.subplot(1,2,2)
    sb.pointplot(data=ride_old, x='boarding_hour', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Hours');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("hourly older people riding pattern: age > 60yrs")

    plt.figure(figsize = [20, 10])
    plt.subplot(2,2,1)
    sb.pointplot(data=ride_old, x='days', y='travelTime_minutes', hue='user_type', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("older people riding pattern throughout the month: age > 60yrs")
      
    plt.subplot(2,2,2)
    sb.pointplot(data=ride_old, x='day_of_week', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-",order=weekday);
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("weekly older people riding pattern based on gender: age > 60yrs")
   
    

oldriders()

def more_from_old_riders():

    plt.figure(figsize = [20,10])
    plt.subplot(3,2,1)
    sb.pointplot(data=ride_old, x='boarding_hour', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Hours');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("hourly older people riding pattern based on gender: age > 60yrs")
    plt.legend(loc=2)

    
    plt.subplot(3,2,2)
    sb.pointplot(data=ride_old, x='days', y='travelTime_minutes', hue='member_gender', dodge=0.3, linestyles="-");
    plt.xlabel('Day of Week');
    plt.ylabel('Avg. Trip Duration in Minute');
    plt.xticks(rotation=45)
    plt.title("older people riding pattern throughout the month based on gender: age > 60yrs")
more_from_old_riders()

• older customers tend to spend far less time riding during weekends on like youth customers but older subcribers spend similar duration of time younger subcribers spend during weekend.
• there seem to be an odd value in the time old occassional customers use of the service at about 2am in the morning but usually these set people use the service from 5am to 9pm for occassional customers while subscribers use it throughout the day where subcribers stay with 20min travel time and occasional users sometimes go up to 60mins.
• on the entire month chart (last chart) where we need insight on how older users use the service. We can see that subscribers averagely spend about 15mins riding except from the 8th-10th but the cycle falls back to normal on the 11th

looking at their riding pattern using gender

• male users tend to use the platform more than other genders and throughout a day on like other genders that start using using the platform from 6am
• female users use the platform consistenly as male users but other gender aren't as consistent.
• other gender spend less time riding from thursday-friday than both male and female genders where male users i.e (sat-sun) looks exactly like that of the younger male users buth female users spend more time riding from wednesay- Friday.

The chart above will prompt more insight to how these use the service during the weekend by looking at the hourly and period features for riders >60yrs

def olderWeekend():
    plt.figure(figsize = [20, 10])
    plt.subplot(1,2,1)
    sb.countplot(data=weekend_ride_old, x='period_of_day',palette="Blues",hue="user_type", order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("Period of day each gender use service the most during weekends for people > 60yrs :Focus(User_type)")
   
    plt.subplot(1,2,2)
    sb.countplot(data=weekend_ride_old, x='period_of_day',palette="Blues",hue="member_gender",order=period);
    plt.xlabel('Period of Day');
    plt.ylabel('Count');
    plt.title("Period of day riders use service the most during weekends for people > 60yrs :Focus(User_type)")
   



    plt.figure(figsize = [20, 10])
    plt.subplot(2,2,1)
    sb.countplot(data=weekend_ride_old, x='boarding_hour',hue="user_type",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("Is there a change in usage of service during weekends for people > 60yrs: focus (User_type)")
    plt.legend(loc=2)


    plt.subplot(2,2,2)
    sb.countplot(data=weekend_ride_old, x='boarding_hour',hue="member_gender",palette="Blues");
    plt.xlabel('Hour of Day');
    plt.ylabel('Count');
    plt.title("What time does each gender make use of the service the most during weekends for people > 60yrs")
    plt.legend(loc=2)
   
    

olderWeekend()

night  = weekend_ride_old[weekend_ride_old["member_gender"] == "other"]
night

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours

0 rows × 25 columns

• though the olders riders are fewer than the younger riders we saw that other genders tend not to come out at night to use the service at all also using percentages on like young riders that rides a slightly more at night older riders use the more in the morning than evening during weekend.
• though we are examining older people from 60 years above we can see that there seem to a normal distributed fig from 6am to 11pm based on gender feature for all genders but even though the user type looks normally distributed yet there are time customers don't use the service at all like 6am-7am and from 8pm till 4am
• Usage of the service for older people peaked at 1pm for subcribers and 10am for customers while male riders rode more at 1pm,3pm then finally peak at 4pm.

	duration_sec	start_time	end_time	start_station_id	start_station_name	start_station_latitude	start_station_longitude	end_station_id	end_station_name	end_station_latitude	...	bike_share_for_all_trip	year	month	days	boarding_hour	period_of_day	day_of_week	age	travelTime_minutes	travelTime_hours

0 rows × 25 columns

Talk about some of the relationships you observed in this part of the investigation. Were there features that strengthened each other in terms of looking at your feature(s) of interest?

user within the ages of 18-40 spend about 10mins on average for both customers and subscribers in 24hrs

• on the weekly chart we can see that the youth trend is similar to that of the whole data where both subcribers and customers spend more travel time riding during the weekends
• throughout the month of february young subcribers spend about 10mins on riding to their destinations but customers are travel time is not constant.

Were there any interesting or surprising interactions between features?

People from 60 years above we can see that there seem to a normal distributed fig from 6am to 11pm based on gender feature for all genders but even though the user type looks normally distributed yet there are time customers don't use the service at all like 6am-7am and from 8pm till 4am

• Usage of the service for older people peaked at 1pm for subcribers and 10am for customers.
• But using gender the peak period is 2pm in the afternoon for all gender.

Conclusions

With the above insight and plots, We can see huge variations of ride durations for the customers but this is not the case for the subscribers.
My interpretation is that customers tend to be tourists or visitors that show up during the day to commute around town so the rides tend to be longer and mostly happen around noon. For the subscribers, it's more likely for them to live in the local area and commute daily depending on their work schedules. That's why there's no massive peak of trip duration time and the durations remain consistent (8am and 17pm) in the both young riders and older riders.

We can also deduce that most riders don't like to share their rides probably due to the fact that riders need to get to where need to get to in time.

The weekend rides that tends to increase

Limitations

There the dataset is enough to enable us build a machine learning solution that can enable us predict users travel_time from station to station. Though there are more but I will limit it to the one stated.

References

• python - How do I get the day of week given a date? - Stack Overflow https://stackoverflow.com/questions/9847213/how-do-i-get-the-day-of-week-given-a-date
• Parts of the Day: Early morning, late morning, etc. | Britannica Dictionary https://www.britannica.com/dictionary/eb/qa/parts-of-the-day-early-morning-late-morning-etc
• python pandas extract year from datetime: df['year'] = df['date'].year is not working - Stack Overflow https://stackoverflow.com/questions/30405413/python-pandas-extract-year-from-datetime-dfyear-dfdate-year-is-not-wo

ride_data.to_csv("new_main.csv")
ride_youth.to_csv("youth_riders.csv")
ride_old.to_csv("old_riders.csv")