In [39]:
# 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
%matplotlib inline
In [3]:
df = pd.read_csv('2017-fordgobike-tripdata.csv')
In [4]:
df.info()
In [5]:
df.dropna(inplace = True)
In [6]:
df.info()
In [7]:
df['start_time'] = df['start_time'].astype('datetime64[ns]')
df['end_time'] = df['end_time'].astype('datetime64[ns]')
In [8]:
df['user_type'] = df['user_type'].astype('category')
df['member_gender'] = df['member_gender'].astype('category')
In [9]:
df.info()
What is the structure of your dataset?¶
- Trip Duration (seconds)
- Start Time and Date
- End Time and Date
- 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 (Subscriber or Customer – “Subscriber” = Member or “Customer” = Casual)
- Member Year of Birth
- Member Gender
What is/are the main feature(s) of interest in your dataset?¶
- I think the ( User_type, member_birth_year, member_gender)
What features in the dataset do you think will help support your investigation into your feature(s) of interest?¶
- User_type
- member_birth_year
- member_gender
- duration_sec
Univariate Exploration¶
In [24]:
# i added the bin edges so that the visual looks good
bin_edges = np.arange(df['member_birth_year'].min(), df['member_birth_year'].max() + 1, 1)
plt.hist(data = df, x ='member_birth_year',bins = bin_edges);
plt.title('The Rider Birth Year');
plt.xlabel('Year of Birth');
plt.ylabel('Counts');
- as you can see here the most riders aged between 29 and 39 years
In [21]:
bin_edges = 10 ** np.arange(1.785, 4.936 + 0.1, 0.1)
ticks = [20,100,200,1000,4000,10000,20000]
labls = ['{}'.format(v) for v in ticks]
plt.hist(data = df, x ='duration_sec', bins = bin_edges);
plt.xscale("log");
plt.xticks(ticks,labls);
plt.title('The Duration in seconds');
plt.xlabel('Duration in Seconds');
plt.ylabel('Counts');
- between 200 and 1000 second is the most time spent on cycling
In [34]:
# i used the color_palettle()[0] to moke the color the original
base_color = sb.color_palette()[0]
sb.countplot(data = df, x = 'member_gender', color = base_color);
plt.title('The Rider Gender');
plt.xlabel('Gender of Rider');
- as you can see here The Number of Males around 350000 Unlike The females, Other.
Discuss the distribution(s) of your variable(s) of interest. Were there any unusual points? Did you need to perform any transformations?¶
- no i guess everything is good!
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 in the duration_sec i had to add ticks and bin edges to make the distribution better
Bivariate Exploration¶
In [35]:
# here i user the "df.query("member_birth_year >= 1950 ")" to Determination of the y axis
sb.violinplot(data = df.query("member_birth_year >= 1950 ")
,x = 'member_gender',y='member_birth_year',color = sb.color_palette()[0]);
plt.title('The Rider Genders and Year of Birth');
plt.xlabel('Gender of Rider');
plt.ylabel('Year of Birth');
- The females are the highest in 1980 and 1990 they are more than the others
In [61]:
q1 = df['duration_sec'].quantile(0.25)
q3 = df['duration_sec'].quantile(0.75)
In [62]:
qr = q3 - q1
qr
Out[62]:
In [63]:
q3
Out[63]:
In [64]:
q1
Out[64]:
- and here i Calculated the first and the third quantile of the Duration to specify the range in the visual
In [36]:
# i used the first and the third quantile here
sb.violinplot(data = df.query('duration_sec <= 838 + 1.5*474'),
x = 'user_type',y='duration_sec',color = sb.color_palette()[0]);
plt.title('The Rider Type and Duration in Seconds');
plt.xlabel('User Type');
plt.ylabel('Duration in Seconds');
- The Coustomers takes more time than the Subscribers
In [37]:
sb.violinplot(data = df.query('duration_sec <= 838 + 1.5*474'),
x = 'member_gender',y='duration_sec',color = sb.color_palette()[0]);
plt.title('The Rider Gender and Duration in Seconds');
plt.xlabel('Gender of Rider');
plt.ylabel('Duration in Seconds');
- as you can see here the males are more than the others with a very small different in The Durations in Seconds
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?¶
as you can see in the first graph i compared the birth year and the gender and it looks like The females are the highest Between 1980 and 1990 they are more than the others
and in the second graph i compared the Duration with the User type and we get that The Coustomers takes more time than the Subscribers
Did you observe any interesting relationships between the other features (not the main feature(s) of interest)?¶
- Yes the Coustomers takes more time than the Subscribers
Multivariate Exploration¶
Duration , Birth Year , User Type¶
In [38]:
x_bin = np.arange( df['member_birth_year'].min(), df['member_birth_year'].max()+5, 5)
y_bin = np.arange(0, (838 + 1.5*474)+500, 500)
_temp = df.sample(300000, random_state=42)
m = sb.FacetGrid(data = _temp, col = 'user_type',height=5.4)
m.map(plt.hist2d, 'member_birth_year','duration_sec',cmin=0.5,cmap = 'viridis_r',bins=[x_bin, y_bin]);
plt.ylim((0,(838 + 1.5*474)));
plt.colorbar();
plt.xlabel('Year of Birth');
plt.ylabel('Duration in Seconds');
- Customer usually takes from 500 to 1000 Seconds in thier Rides
- Subscriber usually takes from 0 to 500 Seconds in thier Rides
- and they are all between 39 and 20 Years old
Duration , Birth Year , Member Gender¶
In [86]:
x_bin = np.arange( df['member_birth_year'].min(), df['member_birth_year'].max()+5, 5)
y_bin = np.arange(0, (838 + 1.5*474)+500, 500)
_temp = df.sample(300000, random_state=42)
m = sb.FacetGrid(data = _temp, col = 'member_gender',height=5.4)
m.map(plt.hist2d, 'member_birth_year','duration_sec',cmin=0.5,cmap = 'viridis_r',bins=[x_bin, y_bin]);
plt.ylim((0,(838 + 1.5*474)));
plt.colorbar();
- Females usually takes from 500 to 1000 Seconds in thier Rides
- Males and Other usually takes from 0 to 1000 Seconds in thier Rides
- and they are all between 39 and 20 Years old
In [19]:
cat_means = df.groupby(['member_gender', 'user_type']).mean()['duration_sec']
cat_means = cat_means.reset_index(name = 'Duration_avg')
cat_means = cat_means.pivot(index = 'user_type', columns = 'member_gender',
values = 'Duration_avg')
sb.heatmap(cat_means, annot = True, fmt = '.3f',
cbar_kws = {'label' : 'mean(Duration_avg)'});
- as you can see the coustomer females are the highest in the average Duration ( top left )
- where the male Subscriber are the lowest in the average Duration ( bottom middle )
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?¶
- You Can see that most of the riders are in between 1980 and 2000
Were there any interesting or surprising interactions between features?¶
- i dont know for sure but i think the duration and the other gender have some interaction.