数据科学生命周期_数据科学项目生命周期第1部分

数据科学生命周期

This is series of how to developed data science project.

这是如何开发数据科学项目的系列。

This is part 1.

这是第1部分。

All the Life-cycle In A Data Science Projects-1. Data Analysis and visualization.2. Feature Engineering.3. Feature Selection.4. Model Building.5. Model Deployment.

数据科学项目中的所有生命周期-1 .数据分析和可视化2。 特征工程3。 功能选择4。 建立模型5。 模型部署。

This whole Data science project life cycle divided into 4 parts.

整个数据科学项目生命周期分为4个部分。

Part 1: How to analysis and visualize the data.

第1部分:如何分析和可视化数据。

Part 2: How to perform feature engineering on data.

第2部分:如何对数据执行特征工程。

Part 3: How to perform feature extraction.

第3部分:如何执行特征提取。

Part 4: model building and development.

第4部分:模型构建和开发。

This tutorial is helpful for them. who has always confused , how to developed data science projects and what is the process for developing data science projects.

本教程对他们有帮助。 谁总是困惑,如何开发数据科学项目以及开发数据科学项目的过程是什么

So, if you are one of them . so you are at right place.

因此,如果您是其中之一。 所以你来对地方了。

Now, In this whole series. we gonna discuss about complete life cycle of data science projects. we will discuss from Scratch.

现在,在整个系列中。 我们将讨论数据科学项目的完整生命周期。 我们将从头开始进行讨论。

In this Part 1. we will see, how to analysis and visualize the data.

在第1部分中,我们将看到如何分析可视化数据。

So, for all of this, we have to take simple data set of Titanic: Machine Learning from Disaster

因此,对于所有这些,我们必须采用“ 泰坦尼克号:灾难中的机器学习”的简单数据集

So, using this dataset we will see, how complete data science life cycle is work.

因此,使用该数据集,我们将看到完整的数据科学生命周期是如何工作的。

In next future tutorial we will take complex dataset.

在下一个未来的教程中,我们将采用复杂的数据集。

dataset link -https://www.kaggle.com/c/titanic/data

数据集链接 -https://www.kaggle.com/c/titanic/data

In this dataset 891 records with 11 features(columns).

在该数据集中,有11个特征(列)的891条记录。

So, let begin with part 1-Analysis and visualize the data in data science life cycle.

因此,让我们从第1部分开始分析,并可视化数据科学生命周期中的数据。

STEP 1-

第1步-

import pandas as pd
import numpy as np
import matplotlib as plt
%matplotlib inline

Import pandas for perform manipulation on dataset.

导入熊猫以对数据集执行操作

Import numpy to deal with mathematical operations.

导入numpy以处理数学运算

Import matplotlib for visualize the dataset in form of graphs.

导入matplotlib图表的形式直观的数据集。

STEP 2-

第2步-

data=pd.read_csv(‘path.csv’)
data.head(5)

Dataset is like this-

数据集这样的-

Image for post

There is 11 features in dataset.

数据集中有11个要素。

1-PassengerId

1-PassengerId

2-Pclass

2类

3-Name

3名

4-Sex

4性别

5-Age

5岁

6-Sibsp

6麻痹

7-Parch

7月

8-Ticket

8票

9-Fare

9票价

10-cabin

10舱

11-embeded

11嵌入式

ABOUT THE DATA –

关于数据–

VariableDefinitionKeysurvivalSurvival0 = No, 1 = YespclassTicket class1 = 1st, 2 = 2nd, 3 = 3rdsexSexAgeAge in yearssibsp# of siblings / spouses aboard the Titanicparch# of parents / children aboard the TitanicticketTicket numberfarePassenger farecabinCabin numberembarkedPort of EmbarkationC = Cherbourg, Q = Queenstown, S = Southampton

VariableDefinitionKeysurvivalSurvival0 =否,1 = YespclassTicket class1 = 1st,2 = 2nd,3 = 3rdsexSexAgeAge年龄在Titanicparch的同胞/配偶中的同胞/配偶数#TitanicticketTicket的票价/乘客乘母票的票价是南安普敦

可变音符 (Variable Notes)

pclass: A proxy for socio-economic status (SES)1st = Upper2nd = Middle3rd = Lower

pclass :社会经济地位(SES)的代理人1st =上层2nd =中间3rd =下层

age: Age is fractional if less than 1. If the age is estimated, is it in the form of xx.5

age :年龄小于1时是小数。如果估计了年龄,则以xx.5的形式出现

sibsp: The dataset defines family relations in this way…Sibling = brother, sister, stepbrother, stepsisterSpouse = husband, wife (mistresses and fiancés were ignored)

sibsp :数据集通过这种方式定义家庭关系… 兄弟姐妹 =兄弟,姐妹,继兄弟,继母配偶 =丈夫,妻子(情妇和未婚夫被忽略)

parch: The dataset defines family relations in this way…Parent = mother, fatherChild = daughter, son, stepdaughter, stepsonSome children travelled only with a nanny, therefore parch=0 for them.

parch :数据集通过这种方式定义家庭关系… 父母 =母亲,父亲孩子 =女儿,儿子,继女,继子有些孩子只带保姆旅行,因此parch = 0。

STEP 3 -After import the dataset. now time to analysis the data.

步骤3-导入数据集后。 现在该分析数据了。

First, we gonna go check the Null value in dataset.

首先,我们要检查数据集中Null值。

null_count=data.isnull().sum()null_count

In that code-

在该代码中-

Now here, We are checking, how many features contains null values.

现在在这里,我们正在检查多少个功能包含值。

here is list..

这是清单。

Image for post
Output
输出量

STEP 4-

第4步-

null_features=null_count[null_count>0].sort_values(ascending=False)
print(null_features)
null_features=[features for features in data.columns if data[features].isnull().sum()>0]
print(null_features)

In 1st line of code – we are extracting only those features with their null values , who has contains 1 or more than 1 Null values.

在第一行代码中,我们仅提取具有值的特征 ,这些特征包含1个或多个1个 Null值。

In 2nd line of code – print those features, who has contains 1 or more than 1 null values.

在第二行代码中–打印那些包含1个或多个1个空值的要素。

In 3rd line of code – we extracting only those features, who has contains 1 or more than 1 null values using list comprehensions.

在代码的第三行中,我们仅使用列表推导提取那些包含1个或多个1个空值的要素

In 4th line of code -Print those all features.

在代码的第4行-P rint所有这些功能。

Image for post
Output
输出量

We will deal with null values in feature engineering part 2.

我们将在要素工程第2部分中处理空值。

STEP 5-

步骤5

data.shape

Check the shape of the dataset.

检查数据集的形状。

Image for post
Output
输出量

STEP 6 - Check unique values in passenger i’d, name, ticket columns.

第6步-检查乘客编号, 姓名机票栏中的唯一值。

print("{} unique values in passenger id columns ".format(len(data.PassengerId.unique())))
print("{} unique values in Name columns ".format(len(data.Name.unique())))
print("{} unique values in Ticket columns ".format(len(data.Ticket.unique())))
Image for post
Output
输出量

891 unique values in passenger id and name columns

乘客ID名称列中的891个唯一值

681 unique values in ticket columns.

凭单列中有681个唯一值。

So , These three columns will not be do any effect on our prediction. these columns not useful, because all values are unique values .so we are dropping these three columns from the dataset.

因此,这三列将不会对我们的预测产生任何影响。 这些没有用,因为所有值都是唯一值。因此我们从数据集中删除了三列

STEP 7- Drop these three columns

步骤7-删除这三

data=data.drop(['PassengerId','Name','Ticket'],axis=1)

STEP 8-

步骤8

data.shape
Image for post
Output
输出量

Check the shape of dataframe once again .It has 891 *9.

再次检查数据框的形状。它具有891 * 9。

Now, After drop three features .9 columns (features) present in dataset.

现在,删除三个功能。 数据集中存在9列 ( 特征 )。

STEP 9 - In this step, we gonna go check is this dataset is balance or not.

步骤9-在这一步中 ,我们将检查此数据集是否平衡。

If dataset is not balance . it might be reason of bad accuracy, because in dataset. if you have many data points which belong only to single class. so this thing may lead over-fitting problem.

如果数据集不平衡。 这可能是因为数据集中的准确性不佳 。 如果您有许多仅属于单个类的数据点。 所以这东西可能会导致过度拟合的问题。

data["Survived"].value_counts().plot(kind='bar')
Image for post
Output
输出量

Here, Dataset is not proper balance, it’s 60–40 ratio.

这里 ,数据集不是适当的平衡, 它是60–40的比率。

So , it will not be effect so much on our predictions. If it effect on our prediction we will deal with this soon.

因此,它对我们的预测影响不大。 如果它影响我们的预测, 我们将尽快处理。

STEP 10 -

步骤10-

datasett=data.copy
not_survived=data[data['Survived']==0]survived=data[data['Survived']==1]

1st line of code — Copy the dataset in dataset variable. because , we will perform some manipulation on data . so that manipulation does not effect on real data. so that we are copying dataset into datasett variable.

第一行代码—将数据集复制到数据集变量中。 因为,我们将对数据进行一些处理 。 这样操作就不会影响真实数据。 所以,我们要复制的数据集datasett变量。

2nd line of code - we are dividing dependent feature (“Survived “) into two part-

代码的第二行-我们将依存特征(“幸存的”)分为两部分,

1. extract all non survival data points from the “Survived” column and store into not_survived variable.

1.从“生存”列中提取所有非生存数据点,并将其存储到not_survived变量中。

2. extract all survival data points from the “Survived” column and store into survival variable.

2.从“ Survived”列中提取所有生存数据点,并将其存储到生存变量中。

It is for visualize the dataset.

用于可视化数据集。

STEP 11 - Now , We are checking of null values’s features(columns) dependency with the Survived or non survived peoples .

步骤11-现在 ,我们正在与生存或未生存的人一起检查值的特征 ( ) 依赖性

First of all , right here, we are checking null values ‘s features(columns) dependency with survived people only.

首先,在这里,我们幸存者一起检查null值的功能(列)依赖性

import matplotlib.pyplot as plt
dataset = survived.copy()%matplotlib inline
for features in null_features:
dataset[features] = np.where(dataset[features].isnull(), 1, 0)
dataset.groupby(features)['Survived'].count().plot.bar()
plt.xlabel(features)
plt.ylabel('Survived')
plt.title(features)
plt.show()

1st line of code- Import matplotlib.lib for visualize purpose.

第一行代码-我导入matplotlib.lib以实现可视化目的。

2nd line of code- copy the only survived people dataset into dataset variable for further processing and it copy because any manipulation does not effect on real data frame.

代码的第二行-将唯一幸存的人员数据集复制到数据集变量中以进行进一步处理, 并进行复制,因为任何操作都不会影响实际数据帧。

3rd line of code - Used loop . This loop will take only null value’s features(columns) only.

第三行代码-使用循环 。 此循环将仅采用值的功能(列)。

4th line of code — Null values replace by the 1 or non null values replace by 0. for check the dependency.

代码的第四行- 值替换为1或非空值替换为0 。 用于检查依赖性

Image for post
Graph 1
图1
Image for post
Graph 2
图2
Image for post
Graph 3
图3

Look in above graphs . all 1 values is basically null values and 0 values is not null values.

上面的图表 。 所有1个值基本上都是值,而0个不是空值。

In age’s graph , dependency is low with null value .

在age的图中,依赖性较低,值为null。

In cabin’s graph, dependency is high with null values.

在机舱的图形中,具有空值的依赖性较高。

In embarked’s graph ,there is approximate not any dependency with null values.

在图的图形中,没有近似值的依赖项为空。

STEP 12- Over here, we are checking dependency with not survived peoples.

第12步-在这里, 我们正在检查对尚未幸存的民族的依赖性。

import matplotlib.pyplot as plt
dataset = not_survived.copy()
%matplotlib inline
for features in null_features:
dataset = not_survived.copy()
dataset[features] = np.where(dataset[features].isnull(), 1, 0)
dataset.groupby(features)['Survived'].count().plot.bar()
plt.xlabel(features)
plt.ylabel('Survived')
plt.title(features)
plt.show()
Image for post
Graph 1
图1
Image for post
Graph 2
图2
Image for post
Graph 3
图3

Same like step 7 . but over here we are dealing with non survival people’s.

与步骤7相同。 但在这里,我们正在与非生存者打交道。

STEP 13 -Let find the relationship between all categories variable(features) with survive peoples.

步骤13-让我们找出所有类别变量 (特征)与生存人群之间的关系

num_features=data.columns
for feature in num_features:
if feature != 'Survived' and feature !='Fare' and feature != 'Age':
survived.groupby(feature)['Survived'].count().plot.bar()
plt.xlabel(feature)
plt.ylabel('Survived')
plt.title(feature)
plt.show()

Survived is dependent variable (feature).

生存的是因变量(特征)。

Fare is numeric variable.

票价是数字变量。

Age is also numeric variable.

年龄也是数字变量。

So , we are not considering these three features (columns) .

因此 ,我们不考虑这三个功能 ( )。

Image for post
Image for post
Image for post
Image for post
Image for post
Image for post

These all are relationship between categorical variables(features) with survived peoples.

这些都是类别变量 ( 特征 )与幸存者之间的关系

STEP 14- Let find the relationship between all categories variable with not survived peoples.

步骤14-让所有类别的变量与没有生存的人之间建立关系

for feature in num_features:
if feature != 'Survived' and feature !='Fare' and feature != 'Age':
not_survived.groupby(feature)['Survived'].count().plot.bar()
plt.xlabel(feature)
plt.ylabel('Not Survived')
plt.title(feature)
plt.show()
Image for post
Image for post
Image for post
Image for post
Image for post
Image for post

These all are relationship between categorical variables(features) with not survived peoples.

这些都是没有生存的人的 分类变量 ( 特征 )之间的关系

STEP 15 -Now check with the numeric variables.

步骤15-现在检查数字变量

non_sur_fare_mean=round(not_survived['Fare'].mean())
non_sur_age_mean=round(not_survived['Age'].mean())
print('its not survived people fare averages',non_sur_fare_mean)
print('its not survived people Age averages',non_sur_age_mean)sur_fare_mean=round(survived['Fare'].mean())
sur_age_mean=round(survived['Age'].mean())
print('its survived people fare averages',sur_fare_mean)
print('its survived people Age averages',sur_age_mean)
Image for post

From this , Its clearly show-

由此可见,它清楚地表明:

Those who were above 31 years old had very few chances to escapes.

那些31岁以上的人很少有机会逃脱。

Those who fares was less than 48 had very few chances to escapes.

票价低于48美元的人几乎没有机会逃脱。

STEP 16- Analysis the numeric features. check distribution of the numeric variable(features).

步骤16- 分析 数字特征 。 检查数字变量的 分布 ( 功能 )。

for feature in num_features:
if feature =='Fare' or feature == 'Age':
data[feature].hist(bins=25)
plt.xlabel(feature)
plt.ylabel("Count")
plt.title(feature)
plt.show()
Image for post
Image for post
Image for post
Image for post

Age followed normal distribution, but Fare is not following normal distribution.

年龄遵循正态分布 ,但票价不遵循正态分布

STEP 17- Check the outliers in dataset.

步骤17-检查数据集中离群值

There are many ways to check the outliers.

有许多方法可以检查异常值。

1- Box plot.

1-盒图。

2- Z-score.

2-Z得分。

3- scatter plot

3-散点图

ETC.

等等。

First of all , Find the outliers using scatter plot-

首先,使用散点图找到离群值-

for feature in num_features:
if feature =='Fare' or feature == 'Age':plt.scatter(data[feature],data['Survived'])
plt.xlabel(feature)
plt.ylabel('Survived')
plt.title(feature)
plt.show()
Image for post
Image for post
Image for post
Image for post
explainjatiopn scatter of fare
解释票价分散

In Age less no of outlier.

在年龄中,没有异常值。

but in fare you can see there is lot of outlier in fare.

但是在票价上您可以看到票价中有很多离群值。

STEP 18- Find outlier using box plot-

步骤18- 使用箱形图查找离群值-

Box plot is better way to check outlier in distribution.

箱形图是检查分布中异常值的更好方法。

for feature in num_features:
if feature =='Fare' or feature == 'Age':
data.boxplot(column=feature)
plt.ylabel(feature)
plt.title(feature)
plt.show()
Image for post
box plot og Age
箱形图与年龄
Image for post
explain age box plot
解释年龄盒图
Image for post
box plot of fare
箱票价格
Image for post
Explanation about image
关于图像的说明

Box plot is also show same picture like scatter.

箱形图也显示散点图。

Age has few no of outlier , but in fare has lot of outliers.

年龄很少有异常值,但票价中有很多异常值。

This first part ended here.

第一部分到此结束。

In this part we saw. how, we can analysis and visualize the dataset.

在这一部分中,我们看到了。 如何分析和可视化数据集。

In next part we will see how feature engineering perform.

在下一部分中,我们将了解要素工程的性能。

翻译自: https://medium.com/@singhpuneei/data-science-project-life-cycle-part-1-f8ebc0737d4

数据科学生命周期

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如若转载,请注明出处:http://www.mzph.cn/news/389481.shtml

如若内容造成侵权/违法违规/事实不符,请联系多彩编程网进行投诉反馈email:809451989@qq.com,一经查实,立即删除!

相关文章

Keras框架:VGG网络代码实现

VGG概念: VGG之所以经典,在于它首次将深度学习做得非常“深”,达 到了16-19层,同时,它用了非常“小”的卷积核(3X3)。 网络框架: VGG的结构: 1、一张原始图片被resize…

Keras框架:resent50代码实现

Residual net概念 概念: Residual net(残差网络):将靠前若干层的某一层数据输出直接跳过多层引入到后面数据层的输入 部分。 残差神经单元:假定某段神经网络的输入是x,期望输出是H(x),如果我们直接将输入x传到输出作…

Tensorflow框架:InceptionV3网络概念及实现

卷积神经网络迁移学习-Inception • 有论文依据表明可以保留训练好的inception模型中所有卷积层的参数,只替换最后一层全连接层。在最后 这一层全连接层之前的网络称为瓶颈层。 • 原理:在训练好的inception模型中,因为将瓶颈层的输出再通过…

成为一名真正的数据科学家有多困难

Data Science and Machine Learning are hard sports to play. It’s difficult enough to motivate yourself to sit down and learn some maths, let alone to becoming an expert on the matter.数据科学和机器学习是一项艰巨的运动。 激励自己坐下来学习一些数学知识是非常…

Ubuntu 装机软件

Ubuntu16.04 软件商店闪退打不开 sudo apt-get updatesudo apt-get dist-upgrade# 应该执行一下更新就好,不需要重新安装软件中心 sudo apt-get install –reinstall software-center Ubuntu16.04 深度美化 https://www.jianshu.com/p/4bd2d9b1af41 Ubuntu18.04 美化…

数据分析中的统计概率_了解统计和概率:成为专家数据科学家

数据分析中的统计概率Data Science is a hot topic nowadays. Organizations consider data scientists to be the Crme de la crme. Everyone in the industry is talking about the potential of data science and what data scientists can bring in their BigTech and FinT…

Keras框架:Mobilenet网络代码实现

Mobilenet概念: MobileNet模型是Google针对手机等嵌入式设备提出的一种轻量级的深层神经网络,其使用的核心思想便是depthwise separable convolution。 Mobilenet思想: 通俗地理解就是3x3的卷积核厚度只有一层,然后在输入张量上…

数据驱动开发_开发数据驱动的股票市场投资方法

数据驱动开发Data driven means that your decision are driven by data and not by emotions. This approach can be very useful in stock market investment. Here is a summary of a data driven approach which I have been taking recently数据驱动意味着您的决定是由数据…

前端之sublime text配置

接下来我们来了解如何调整sublime text的配置,可能很多同学下载sublime text的时候就是把它当成记事本来使用,也就是没有做任何自定义的配置,做一些自定义的配置可以让sublime text更适合我们的开发习惯。 那么在利用刚才的命令面板我们怎么打…

python 时间序列预测_使用Python进行动手时间序列预测

python 时间序列预测Time series analysis is the endeavor of extracting meaningful summary and statistical information from data points that are in chronological order. They are widely used in applied science and engineering which involves temporal measureme…

keras框架:目标检测Faster-RCNN思想及代码

Faster-RCNN(RPN CNN ROI)概念 Faster RCNN可以分为4个主要内容: Conv layers:作为一种CNN网络目标检测方法,Faster RCNN首先使用一组基础的convrelupooling层提取 image的feature maps。该feature maps被共享用于…

算法偏见是什么_算法可能会使任何人(包括您)有偏见

算法偏见是什么在上一篇文章中,我们展示了当数据将情绪从动作中剥离时会发生什么 (In the last article, we showed what happens when data strip emotions out of an action) In Part 1 of this series, we argued that data can turn anyone into a psychopath, …

大数据笔记-0907

2019独角兽企业重金招聘Python工程师标准>>> 复习: 1.clear清屏 2.vi vi xxx.log i-->edit esc-->command shift:-->end 输入 wq 3.cat xxx.log 查看 --------------------------- 1.pwd 查看当前光标所在的path 2.家目录 /boot swap / 根目录 起始位置 家…

Tensorflow框架:目标检测Yolo思想

Yolo-You Only Look Once YOLO算法采用一个单独的CNN模型实现end-to-end的目标检测: Resize成448448,图片分割得到77网格(cell)CNN提取特征和预测:卷积部分负责提取特征。全链接部分负责预测:过滤bbox(通过nms&#…

线性回归非线性回归_了解线性回归

线性回归非线性回归Let’s say you’re looking to buy a new PC from an online store (and you’re most interested in how much RAM it has) and you see on their first page some PCs with 4GB at $100, then some with 16 GB at $1000. Your budget is $500. So, you es…

朴素贝叶斯和贝叶斯估计_贝叶斯估计收入增长的方法

朴素贝叶斯和贝叶斯估计Note from Towards Data Science’s editors: While we allow independent authors to publish articles in accordance with our rules and guidelines, we do not endorse each author’s contribution. You should not rely on an author’s works wi…

numpy统计分布显示

import numpy as np from sklearn.datasets import load_iris dataload_iris()petal_lengthnumpy.array(list(len[2]for len in data[data]))#取出花瓣长度数据 print(np.max(petal_length))#花瓣长度最大值 print(np.mean(petal_length))#花瓣长度平均值 print(np.std(petal_l…

Keras框架:人脸检测-mtcnn思想及代码

人脸检测-mtcnn 概念: MTCNN,英文全称是Multi-task convolutional neural network,中文全称是多任务卷积神经网络, 该神经网络将人脸区域检测与人脸关键点检测放在了一起。 从工程实践上,MTCNN是一种检测速度和准确率…

python中格式化字符串_Python中所有字符串格式化的指南

python中格式化字符串Strings are one of the most essential and used datatypes in programming. It allows the computer to interact and communicate with the world, such as printing instructions or reading input from the user. The ability to manipulate and form…

Javassist实现JDK动态代理

提到JDK动态代理,相信很多人并不陌生。然而,对于动态代理的实现原理,以及如何编码实现动态代理功能,可能知道的人就比较少了。接下一来,我们就一起来看看JDK动态代理的基本原理,以及如何通过Javassist进行模…