数据集与探索性数据分析

1import pandas as pd; import numpy as np
2import matplotlib.pyplot as plt
3import seaborn as sns
4sns.set_style('whitegrid') # just optional!
5
6train = pd.read_csv('train_titanic_Xy.csv')
7# Always good to see how the data look like, head() is useful!
8print(train.head(2))

1print(train.info()) # try ==> train.info(verbose=False)

1#round((train.isnull().sum())/(train.isnull().count())*100,1)
2pct_missing = round((train.isnull().sum())/(train.isnull().count())*100,1)
3print(pct_missing.sort_values(ascending=False).head())

1# heatmap using seaborn, you can set the figure size if you want!
2plt.figure(figsize = (14,7))
3sns.heatmap(data = train.isnull());

1plt.figure(figsize = (14,7))
2sns.heatmap(data=train.isnull(), yticklabels=False, cbar=False, cmap='viridis' );

1plt.figure(figsize = (14,7))
2sns.countplot(x='Survived', data=train);# try different palette, such as 'coolwarm' or anyother!

1plt.figure(figsize = (14,7))
2sns.countplot(x='Survived', hue='Sex',data=train, palette='coolwarm'); # palette is another useful parameter!

1# Let's check the number of classes again to re-confirm!
2print(train['Pclass'].unique())

1plt.figure(figsize = (14,7))
2sns.countplot(x='Survived', hue='Pclass',data=train);
3plt.legend(loc=1);

1# hue is useful here!
2plt.figure(figsize = (14,7))
3sns.countplot(x='Embarked',data=train, hue='Survived');#hue='Pclass')
4plt.legend(loc=1);

1print("Survival rate with respect to the port of embarkation:")
2print("\tS: {} %".format(train[(train.Embarked == 'S') & (train.Survived==1)
3                ]['Survived'].count()/len(train[(train.Embarked == 'S')])*100))
4print("\tC: {} %".format(train[(train.Embarked == 'C') & (train.Survived==1)
5                ]['Survived'].count()/len(train[(train.Embarked == 'C')])*100))
6print("\tQ: {} %".format(train[(train.Embarked == 'Q') & (train.Survived==1)
7                ]['Survived'].count()/len(train[(train.Embarked == 'Q')])*100))

1plt.figure(figsize = (14,7))
2sns.countplot(x='Embarked',data=train, hue='Pclass')
3plt.legend(loc=1);

1# Parch -- no. of parents / children aboard the Titanic
2plt.figure(figsize = (14,7))
3sns.countplot(x='Parch',data=train, hue='Survived')
4plt.legend(loc=1);

1plt.figure(figsize = (14,7))
2sns.distplot(train['Age'].dropna(),kde=False,color='green',bins=30); # try with dropna(), what do you see?
3#train['Age'].hist(bins=30,color='green',alpha=0.5) # using pandas data visualizations