Python数据分析流程

一.数据分析的步骤:

1.查看数据并提出问题

2.数据清洗

3.代码编写,提取出结果数据,并分析是否有异常数据,修改代码

4.根据数据选择合适的图表进行展示

5.根据图表小组讨论交流获得最终的结果

二.环境与原始数据准备

安装Anaconda2版本,同时更新软件包更新最新版本  conda upgrade --all

下载first.zip文件,解压

里面有3张csv文件分别是enrollments.csv,daily_engagements.csv,project_submission.csv和一个ipython的notebook

启动cmd 切换到解压之后的文件 输入 jupyter notebook 启动ipython笔记本

三.分析数据

1.从csv加载数据

import unicodecsv


def readcsv(filename):
    with open(filename,'rb') as f:
        #以字典的形式存放每一行数据
        reader = unicodecsv.DictReader(f)
        return list(reader)    

## 从 daily_engagement.csv 和 project_submissions.csv 载入数据并存
## 储至下面的变量中，然后检查每张表的第1行。

daily_engagement = readcsv('daily-engagement.csv')
project_submissions = readcsv('project-submissions.csv')
enrollments = readcsv('enrollments.csv')

print daily_engagement[0]
print project_submissions[0]
print enrollments[0]

 2.修正数据类型

from datetime import datetime as dt

# 将字符串格式的时间转为 Python datetime 类型的时间。
# 如果没有时间字符串传入，返回 None
def parse_date(date):
    if date == '':
        return None
    else:
        return dt.strptime(date, '%Y-%m-%d')

    
# 将可能是空字符串或字符串类型的数据转为 整型 或 None。
def parse_maybe_int(i):
    if i == '':
        return None
    else:
        return int(i)

# 清理 enrollments 表格中的数据类型(取消的日期,参加日期,退出的天数,是否取消,是否是Udacity测试账号)
for enrollment in enrollments:
    enrollment['cancel_date'] = parse_date(enrollment['cancel_date'])
    enrollment['join_date'] = parse_date(enrollment['join_date'])
    enrollment['days_to_cancel'] = parse_maybe_int(enrollment['days_to_cancel'])
    enrollment['is_canceled'] = enrollment['is_canceled'] == 'True'
    enrollment['is_udacity'] = enrollment['is_udacity'] == 'True'
    
enrollments[0]

# 清理 engagement 的数据类型(时间,课程数量,课程完成数量,项目完成情况,共花费多少时间)
for engagement_record in daily_engagement:
    engagement_record['utc_date'] = parse_date(engagement_record['utc_date'])
    engagement_record['num_courses_visited'] = int(float(engagement_record['num_courses_visited']))
    engagement_record['lessons_completed'] = int(float(engagement_record['lessons_completed']))
    engagement_record['projects_completed'] = int(float(engagement_record['projects_completed']))
    engagement_record['total_minutes_visited'] = float(engagement_record['total_minutes_visited'])
    
daily_engagement[0]

# 清理 submissions 的数据类型(项目创建的时间,完成的时间)
for submission in project_submissions:
    submission['creation_date'] = parse_date(submission['creation_date'])
    submission['completion_date'] = parse_date(submission['completion_date'])

project_submissions[0]

3.修改数据中的格式问题

## 将 daily_engagement 表中的 "acct" 重命名为 ”account_key"
for engagement_record in daily_engagement:
    engagement_record['account_key'] = engagement_record['acct']
    del [engagement_record['acct']]

 4.探索数据

## 计算每张表中的总行数，和独立学生（拥有独立的 account keys）的数量
def unique_student_data(data):
    unique_data = set()
    for data_point in data:
        unique_data.add(data_point['account_key'])
    return unique_data
len(enrollments)
unique_enrolled_students = unique_student_data(enrollments)
len(unique_enrolled_students)

len(daily_engagement)
unique_daily_engagement = unique_student_data(daily_engagement)
len(unique_daily_engagement)

len(project_submissions)
unique_project_submissions = unique_student_data(project_submissions)
len(unique_project_submissions)

 5.找出问题数据

## 计算出有问题的数据点条数（在 enrollments 中存在，但在 engagement 表中缺失）
num_problem_students = 0
for enrollment in enrollments:
    if enrollment['account_key'] not in unique_daily_engagement and enrollment['join_date'] != enrollment['cancel_date']:
        num_problem_students +=1
        print enrollment
        print num_problem_students

 6.追踪剩余的问题(移除数据集的测试账号)

# 为所有 Udacity 测试帐号建立一组 set 
udacity_test_account = set()
for enrollment in enrollments:
    if enrollment['is_udacity']:
        udacity_test_account.add(enrollment['account_key'])
len(udacity_test_account)


# 通过 account_key 删除所有 Udacity 的测试帐号
def remove_udacity_account(data):
    non_udacity_data = []
    for data_point in data:
        if data_point['account_key'] not in udacity_test_account:
            non_udacity_data.append(data_point)
    return non_udacity_data

# 从3张表中移除所有 Udacity 的测试帐号
non_udacity_enrollments = remove_udacity_account(enrollments)
non_udacity_engagement = remove_udacity_account(daily_engagement)
non_udacity_submissions = remove_udacity_account(project_submissions)

#创建一个叫 paid_students 的字典，并在字典中存储所有还没有取消或者注册时间超过7天的学生
paid_students = {}
for enrollment in non_udacity_enrollments:
　　#如果没有取消并且退课的期限已经超过,就记录学生的key和报名时间
    if not enrollment['is_canceled'] or enrollment['days_to_cancel'] > 7:
        account_key = enrollment['account_key']
        enrollment_date = enrollment['join_date']
　　　　 #如果account_key不在已缴费的记录中,则将学生记录添加进paid_student中
        if account_key not in paid_students or enrollment_date > paid_students[account_key]:
            paid_students[account_key] = enrollment_date
len(paid_students)#获取了所有已入学的学生记录

 7.获取第一周就已经付费报名的学生

#计算时间差,一周以内,按天计算
def within_one_week(join_date ,engagement_date):
    time_delta = join_date - enrollment_date
    return time_delta.days >= 0 and time_delta.days < 7
    


#存放已报名的用户
def remove_free_trial_cancels(data):
    new_data = []
    for data_point in data:
        if data_point['account_key'] in paid_students:
            new_data.append(data_point)
    return new_data


paid_enrollment = remove_free_trial_cancels(non_udacity_enrollments)
paid_engagement = remove_free_trial_cancels(non_udacity_engagement)
paid_project_missions = remove_free_trial_cancels(non_udacity_submissions)

print len(paid_enrollment)
print len(paid_engagement)
print len(paid_project_missions)

## 创建一个 engagement 记录的列表，该列表只包括付费学生以及加入的前7天的学生的记录 
## 输入符合要求的行数
paid_engagement_in_first_week = []
for engagement_record in paid_engagement:
    join_date = paid_students[engagement_record['account_key']]
    engagement_record_date = engagement_record['utc_date']
    if within_one_week(join_date,engagement_record_date):
        paid_engagement_in_first_week.append(engagement_record)
len(paid_engagement_in_first_week)

from collections import defaultdict
import numpy as np
#创建基于 student 对 engagement 进行分组的字典，字典的键为帐号（account key），值为包含互动记录的列表
def group_data(data,key_name):
    grouped_data = defaultdict(list)
    for data_point in data:
        key = data_point[key_name]
        grouped_data[key].append(data_point)
    return grouped_data

# 创建一个包含学生在第1周在教室所花总时间和字典。键为帐号（account key），值为数字（所花总时间）
def sum_grouped_items(grouped_data,field_name):
    sumed_data = {}
    for key,data_points in grouped_data.items():
        total = 0
        for data_point in data_points:
            total += data_point[field_name]
        sumed_data[key] = total
    return sumed_data

# 汇总和描述关于教室所花时间的数据
def describe_data(data):
    print 'Mean:', np.mean(data)
    print 'Standard deviation:', np.std(data)
    print 'Minimum:', np.min(data)
    print 'Maximum:', np.max(data)

 8.获取学习时间最长的学生和时间

total_minutes_by_account = sum_grouped_items(engagement_by_account,'total_minutes_visited')

student_with_max_minutes = None
max_minutes = 0
for student,total_nums in total_minutes_by_account.items():
    if total_nums > max_minutes:
        max_minutes = total_nums
        student_with_max_minutes = student
print max_minutes

for engagement_record in paid_engagement_in_first_week:
    if engagement_record['account_key'] == student:
        print engagement_record

 9.找出第一周的访问数

## 找出第1周学生访问教室天数的平均值、标准差、最小值、最大值。
for engagement_record in paid_engagement:
    if engagement_record['num_courses_visited'] > 0:
        engagement_record['has_visited'] = 1
    else:
        engagement_record['has_visited'] = 0
        
days_visited_by_account = sum_grouped_items(engagement_by_account,'has_visited')
describe_data(days_visited_by_account.values())

 10.区分项目通过的学生

## 创建两个付费学生第1周的互动数据列表（engagement）。第1个包含通过项目的学生，第2个包含没通过项目的学生。

subway_project_lesson_keys = ['746169184', '3176718735']
#定义存放通过项目的学员的key
pass_subway_project = set()
for submission in paid_project_missions:
    project = submission['lesson_key']
    rating = submission['assigned_rating']
　　#如果等级是passed和distinction加入到pass_subway_project集合中
    if project in subway_project_lesson_keys and (rating == 'PASSED' or rating == 'DISTINCTION'):
        pass_subway_project.add(submission['account_key'])

passing_engagement = [] #存放通过项目的学生
non_passing_engagement =[] #存放没有通过项目的学生

for engagement_record in paid_engagement_in_first_week:
    if engagement_record['account_key'] in pass_subway_project:
        passing_engagement.append(engagement_record)
    else:
        non_passing_engagement.append(engagement_record)
        
print len(passing_engagement)
print len(non_passing_engagement)

 11.对比两组学生的数据

## 计算你所感兴趣的数据指标，并分析通过项目和没有通过项目的两组学生有何异同。
## 你可以从我们之前使用过的数据指标开始（教室的访问时间、课程完成数、访问天数）。
passing_engagement_by_account = group_data(passing_engagement,'account_key')
non_passing_engagement_by_account = group_data(non_passing_engagement,'account_key')

print 'non-passing students'
non_passing_minute = sum_grouped_items(non_passing_engagement_by_account,'total_minutes_visited')
describe_data(non_passing_minute.values())
print 'passing students'
passing_minute = sum_grouped_items(passing_engagement_by_account,'total_minutes_visited')
describe_data(passing_minute.values())

print 'non-passing lessons'
non_passing_lessons = sum_grouped_items(non_passing_engagement_by_account,'lessons_completed')
describe_data(non_passing_lessons.values())
print 'passing lessons'
passing_lessons = sum_grouped_items(passing_engagement_by_account,'lessons_completed')
describe_data(passing_lessons.values())

print 'non-passing visited'
non_passing_visited = sum_grouped_items(non_passing_engagement_by_account,'has_visited')
describe_data(non_passing_visited.values())
print 'passing visited'
passing_visited = sum_grouped_items(passing_engagement_by_account,'has_visited')
describe_data(passing_visited.values())

 12.绘制直方图

%pylab inline
import matplotlib.pyplot as plt
import numpy as np

def describe_data(data):
    print 'Mean:', np.mean(data)
    print 'Standard deviation:', np.std(data)
    print 'Minimum:', np.min(data)
    print 'Maximum:', np.max(data)
    plt.hist(data)
    
describe_data(passing_minute.values())
describe_data(non_passing_minute.values())

 13.改进图表并分析

## 至少改进一幅之前的可视化图表，尝试导入 seaborn 库使你的图表看起来更美观。
## 加入轴标签及表头，并修改一个或多个 hist() 内的变量。
%pylab inline
import seaborn as sns
sns.set(color_codes=True)
plt.hist(non_passing_minute.values(),bins=8)
plt.xlabel('mean of minut')
plt.title('Distribution of classroom visits in the first week ' + 
          'for students who do not pass the subway project')

plt.hist(passing_minute.values(),bins=8)
plt.xlabel('mean of minut')
plt.title('Distribution of classroom visits in the first week ' + 
          'for students who do not pass the subway project')