Task 1 Titanic top 4% with ensemble modeling

One 、 About python Of nan

NaN（not a number）, In mathematical representation, a number that cannot be represented , There is usually another expression here inf,inf and nan The difference is ,inf Is a floating-point number that exceeds the range of floating-point representation （ Its essence is still a number , It's just that he's infinite , Therefore, floating point numbers cannot be used to represent , such as 1/0）, and nan It generally represents a non floating point number （ For example, irrational numbers ）.

python Medium nan What does that mean? - You know

Two 、 Using the map base Tukey method Detect outliers in the data set

Using the map base Tukey method Detect outliers in the data set _vincent_duan The blog of -CSDN Blog _tukey Method

In a dataset, if an observation is unusually larger or smaller than other data in the dataset , We call it a suspected outlier . The existence of suspected outliers , It will have an inappropriate impact on the subsequent calculation results , It is very necessary to detect suspected abnormal values and deal with them properly .

A classic method of calculating suspected outliers in data sets is Tukey method. This method first calculates the quartile of the data set （Q1） And the third quartile （Q3）, So as to calculate the interquartile spacing （IQR）, Then it will be less than Q1 - 1.5IQR Or greater than Q3 + 1.5IQR Data points of are regarded as suspected outliers . We can use this method in DataFrame Detect outliers in .

Tukey‘s test Method outliers _weixin_30466421 The blog of -CSDN Blog

1.np.percentile() function

np.percentile() Function super detailed _ Kana likes fish blog -CSDN Blog _np.percentile()

2.Python collections.Counter() usage

Python collections.Counter() usage _pvop The blog of -CSDN Blog _collections.counter python

3.python_4.loc() and iloc(） function

python_4.loc() and iloc(） function _Pepei. The blog of -CSDN Blog _loc()

# Drop outliers
train = train.drop(Outliers_to_drop, axis = 0).reset_index(drop=True)# Note that here is the deletion line ,axis = 0

4.reset_index(drop=True)

reset_index(drop=True)_Yukyin The blog of -CSDN Blog

python in axis=0 axis=1 The understanding of the _《 All night 》 The blog of -CSDN Blog _python Of axis=1

5.fillna()

fillna() Function details _Denver_Liao The blog of -CSDN Blog _.fillna

6.info() Function and describe function

info() Function and describe function _ Yi Shang's blog -CSDN Blog _info() function

7.head() function

Pandas Medium head( ) function _ac Student's blog -CSDN Blog _head function

3、 ... and 、 Characteristics analysis

1. Numerical characteristics

1.1.corr() function as well as Heat map heatmap

corr() Function usage _ Study hard --ing The blog of -CSDN Blog _corr

pandas Medium data.corr() Function method description and use _ A cute new blog -CSDN Blog _data.corr()

【 Drawing topic 】sns.heatmap Introduction to the usage of _DrCrypto The blog of -CSDN Blog _sns.heatmap()

1.2. catplot() function

Pycharm Report errors ：FutureWarning: `distplot` is a deprecated function and will be removed in a future version_lith__romantic The blog of -CSDN Blog

Detailed data visualization Library Seaborn course ( Two )——catplot： Draw the coordinate axis with classified data _pingzishinee The blog of -CSDN Blog _catplot

1.3.despine() function  

seaborn despine() Use of functions _ Boulder's blog -CSDN Blog

2. Category features

Four 、 Feature Engineering