Python for Time Series Data Analysis (Udemy) 1

UdemyでPython for TIme Series Data Analysisのクラスの学習を進めている。Pythonで機械学習を用いて時系列データの予測方法が解説される。まずはPythonの使い方で、numpyとpandasの基礎を学習した。

numpyの学習コード1つ目

import numpy as np
mylist = [1,2,3]
type(mylist)
np.array(mylist)
arr = np.array(mylist)
arr
mylist = [[1,2,3],[4,5,6],[7,8,9]]
mylist
mymatrix = np.array(mylist)
mymatrix.shape
mynewmatrix = np.array(mylist)
mynewmatrix = np.array

np.arange(0,10,2)
np.zeros((4,10))

np.ones((5,5)) + 4
np.ones(4)*100
[1,1,1,1]*100
np.linspace(0,10)
np.eye(5)

np.random.rand(5,5)
np.random.randn(10)
np.random.randint(1,100,10)
np.random.seed(555)
np.random.rand(4)
arr = np.arange(25)
arr
ranarr = np.random.randint(0,50,10)
ranarr
arr.reshape(5,5)
ranarr.max()
ranarr.argmax()
ranarr.argmin()

numpyの学習コード2つ目

import numpy as np
arr = np.arange(0,11)
arr[8]
arr[1:5]
arr[0:5]
arr[5:]
arr + 100
new_arr = arr/2
new_arr
arr**2
arr
slice_of_arr = arr[0:6]
slice_of_arr
slice_of_arr[:] = 99
slice_of_arr
arr
arr_2d = np.array([[5,10,15],[20,25,30],[35,40,45]])
arr_2d.shape
arr_2d[1,1]
arr_2d[2,2]
arr_2d
arr_2d[:2,1:]
arr = np.arange(1,11)
arr
bool_arr = arr > 4
arr[bool_arr]
arr[arr>4]
arr[arr<=6]

numpyの学習コード3つ目

import numpy as np
arr = np.arange(0,10)
arr
arr + 100
arr / 2
arr**2
(arr+2)/100
arr
arr + arr
1/arr
arr/arr
np.sqrt(arr)
np.log(arr)
np.sin(arr)
arr.sum()
arr.mean()
arr.max()
arr_2d = np.array([[1,2,3,4],[5,6,7,8],[9,10,11,12]])
arr_2d
arr_2d.shape
arr_2d.sum()
arr_2d.sum(axis=0)

numpyの学習コード4つ目

import numpy as np
np.zeros(10)
np.ones(10)
np.ones(10)*5
np.arange(10,51)
np.arange(10,51,2)
np.arange(9).reshape(3,3)
np.eye(3,3)
np.random.rand(1)
np.random.randn(25)
arr = np.arange(1,101)/100
arr.reshape(10,10)
np.linspace(0,1,20)
mat = np.arange(1,26).reshape(5,5)
mat
mat[2:,1:]
mat[3,4]
mat[:3,1:2]
mat[4,:]
mat[3:5,:]
mat.sum()
mat.std()
mat.sum(axis=0)
np.random.seed(101)
np.random.rand(1)

pandasの学習コード1つ目

import numpy as np
import pandas as pd
labels = ['a','b','c']
mylist = [10,20,30]
arr = np.array(mylist)
arr
d = {'a':10,'b':20,'c':30}
pd.Series(data=mylist)
pd.Series(arr,index=labels)
pd.Series(data=['d','a','e'])
ser1 = pd.Series([1,2,3,4],index=['USA','Germany','USSR','Japan'])
ser1
ser1['USA']
ser2 = pd.Series([1,4,5,6],index=['USA','Germany','Italy','Japan'])
ser2
ser1+ser2

pandasの学習コード2つ目

import pandas as pd
import numpy as np
from numpy.random import randn
np.random.seed(101)
rand_mat = randn(5,4)
rand_mat
df = pd.DataFrame(data=rand_mat,index='A B C D E'.split(),columns='W X Y Z'.split())
df
df[['W','Y']]
df['NEW'] = df['W'] + df['Y']
df
df.drop('NEW',axis=1,inplace=True)
df
df.loc['A']
df.iloc[2]
df
df_bool = df>0
df[df_bool]
df[df['W'] > 0]
cond1 = df['W'] > 0
cond2 = df['Y'] > 1
df[(cond1) & (cond2)]
df
df.reset_index()
new_ind = 'CA NY WY OR CO'.split()
new_ind
df['States'] = new_ind
df.set_index('States',inplace=True)
ser_w = df['W'] > 0

pandasの学習コード3つ目

import numpy as np
import pandas as pd
df = pd.DataFrame({'A':[1,2,np.nan],'B':[5,np.nan,np.nan],'C':[1,2,3]})
df
df.dropna()
df.dropna(axis=1)
df.dropna(thresh=2)
df.fillna(value='FILL VALUE')
df.mean()
df.fillna(df.mean())
df['A'].fillna(value=df['A'].mean())