Python 股票分析入門

繆宇發表於2018-11-06

初入資料分析大門,感覺需要補的知識太多太多。不太建議系統補齊各種知識,因為大概率會倒在半路上。

以專案為導向,梳理大概流程,對流程中所需的知識點進行大致學習,以後遇到知識盲點再回來補。

接下來我以股票分析為例來了解資料分析流程。

個人認為股票分析無非就分為獲取資料資料處理資料視覺化三個部分,依賴 Python 強大的第三方開源庫,上手難度變得非常低。

所需知識

  • 基礎
    • Python
  • 環境
    • anaconda
  • 資料獲取
    • pandas_datareader
  • 資料處理
    • numpy
    • pandas
  • 資料視覺化
    • matplotlib seaborn

目的

  • 分析股票走勢
  • 多隻股票的關係

Python 基本語法

推薦去看廖雪峰的 Python 教程,淺顯易懂,上手很快。

安裝 anaconda

Anaconda是一個包含180+的科學包及其依賴項的開源Python發行版本。

安裝 anaconda,直接去anaconda 官網下載安裝即可。

anaconda 安裝成功後,會自帶安裝 Jupyter,jupyter 主要用於我們程式碼的編寫和執行。

建立一個新資料夾 stock-market-analysis ,進入當前目錄,啟動 jupyter 。

# 啟動 jupyter
conda notebook 
複製程式碼

啟動成功,在瀏覽器中開啟 http://localhost:8888/tree) ,單擊 new,建立一個新的 notebook 就可以開始愉快的玩耍了!

numpy

numpy 是一個用於科學計算的 Python 庫。

基本用法

# 引入 numpy
import numpy as np
複製程式碼
# 建立一個長度為15,3乘5的二維陣列
a = np.arange(15).reshape(3, 5)
# 列印a
a
複製程式碼
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])
複製程式碼
# 建立一個長度為15,間隔10,3乘5的二維陣列
b = np.arange( 1, 150, 10 ).reshape(3, 5)
# 列印b
b
複製程式碼
array([[  1,  11,  21,  31,  41],
       [ 51,  61,  71,  81,  91],
       [101, 111, 121, 131, 141]])
複製程式碼
# 兩個二維陣列相加 
a + b
複製程式碼
array([[  1,  12,  23,  34,  45],
       [ 56,  67,  78,  89, 100],
       [111, 122, 133, 144, 155]])
複製程式碼

去官網瞭解更多

pandas

pandas 是一個基於 numpy 強大的 Python 資料分析包,它提供了很高階的資料結構和大量處理資料的方法。最終目的是為了我們更好的理解和處理資料。

基本用法

pandas 提供了兩種資料結構,Series 和 DataFrame。

import numpy as np
import pandas as pd
from pandas import Series, DataFrame
複製程式碼

Series

Series 類似於字典,可以根據索引查詢對應值

# 建立一個長度為4,1到10的隨機整數
Series(np.random.randint(1,10,4))
複製程式碼
0    4
1    2
2    3
3    1
dtype: int64
複製程式碼
# 指定index
s = Series(np.random.randint(1,10,4), index=['a','b','c','d'])
s
複製程式碼
a    4
b    5
c    5
d    8
dtype: int64
複製程式碼
# 根據 index 查詢
s['a']
複製程式碼
4
複製程式碼

DataFrame

DataFrame 是二維的資料結構,可以用行列的方式表示,可以把它想象成一個 Excel 表。

# 生成值為時間的陣列
dates = pd.date_range('20130101', periods=6)
dates
複製程式碼
DatetimeIndex(['2013-01-01', '2013-01-02', '2013-01-03', '2013-01-04',
               '2013-01-05', '2013-01-06'],
              dtype='datetime64[ns]', freq='D')
複製程式碼
# 生成 index 為日期,列名為 ABCD 的 DataFrame
df = DataFrame(np.random.randint(1,10,24).reshape(6,4), index=dates, columns=list('ABCD'))
df
複製程式碼
A B C D
2013-01-01 4 3 5 3
2013-01-02 3 1 1 8
2013-01-03 6 1 8 6
2013-01-04 8 8 9 2
2013-01-05 1 5 1 8
2013-01-06 2 4 9 5

去官網瞭解更多

matplotlib seaborn

matplotlib 是 Python 非常重要的資料視覺化庫,而 seaborn 是基於 matplotlib 開發的視覺化庫,更為強大易用。

基本用法

import numpy as np
import pandas as pd
from pandas import Series, DataFrame
import matplotlib.pyplot as plt
import seaborn as sns
複製程式碼
# 建立一個 Series,1000個從1到100間隔均勻的陣列
s = Series(np.linspace(1, 100, 1000))
複製程式碼
s.plot()
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x1a226a23c8>
複製程式碼

Python 股票分析入門

# 建立一個 DataFrame,1到10的隨機整數,10乘4的二維陣列,列名為 a,b,c,d
df = DataFrame(np.random.randint(1,10,40).reshape(10,4),columns=list("abcd"))
複製程式碼
df
複製程式碼
a b c d
0 3 6 8 9
1 2 6 9 9
2 8 9 2 9
3 7 4 5 4
4 7 9 5 8
5 9 9 3 3
6 5 2 3 1
7 1 8 3 7
8 3 1 7 2
9 2 1 9 7
# matpoltlib 畫圖
df.plot()
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x11fe2ab38>
複製程式碼

Python 股票分析入門

# 對每列進行求和
df_sum = df.sum()
# 指定柱狀圖
df_sum.plot(kind='bar')
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x1a22821668>
複製程式碼

Python 股票分析入門

# seaborn 畫圖
index = df_sum.index
values = df_sum.values
sns.barplot(index, values)
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x1a229041d0>
複製程式碼

Python 股票分析入門

去 matplotlib 官網瞭解更多

去 seaborn 官網瞭解更多

有了以上的一些基本知識就可開始分析股票了。

股票分析

股票分析的步驟:

  • 獲取資料
  • 處理資料
  • 資料視覺化
  • 分析資料
# basic
import numpy as np
import pandas as pd
from pandas import Series, DataFrame

# get data
import pandas_datareader as pdr

# visual
import matplotlib.pyplot as plt
import seaborn as sns

# time
from datetime import datetime
複製程式碼
# pandas_datareader 這個庫提供 API 來獲取股票資料
# get_data_yahoo 代表資料來源來自 yahoo,'BABA' 是阿里巴巴的股票程式碼
df = pdr.get_data_yahoo('BABA')
# 由於資料較多,我們只取頭部的五條資料來看
# high 表示最高價,low 表示最低價,open 表示開盤價,close 表示收盤價,volume 表示交易量
df.head()
複製程式碼
High Low Open Close Volume Adj Close
Date
2014-09-19 99.699997 89.949997 92.699997 93.889999 271879400 93.889999
2014-09-22 92.949997 89.500000 92.699997 89.889999 66657800 89.889999
2014-09-23 90.480003 86.620003 88.940002 87.169998 39009800 87.169998
2014-09-24 90.570000 87.220001 88.470001 90.570000 32088000 90.570000
2014-09-25 91.500000 88.500000 91.089996 88.919998 28598000 88.919998
# 我們取五家從2015年1月1號至今的股票資料進行分析
# 'MSFT':微軟, 'AAPL':蘋果, 'AMZN':亞馬遜, 'FB':臉書, 'GOOG':谷歌
start = datetime(2015,1,1)
company = ['MSFT', 'AAPL', 'AMZN', 'FB', 'GOOG']
top_df = pdr.get_data_yahoo(company, start=start)
複製程式碼
# 得到一個資料集合,這個資料集合是 pandas.core 資料結構
type(top_df)
複製程式碼
pandas.core.frame.DataFrame
複製程式碼
# 看一下他們最近幾日的收盤價
top_df['Close'].tail()
複製程式碼
Symbols AAPL AMZN FB GOOG MSFT
Date
2018-10-30 213.300003 1530.420044 146.220001 1036.209961 103.730003
2018-10-31 218.860001 1598.010010 151.789993 1076.770020 106.809998
2018-11-01 222.220001 1665.530029 151.750000 1070.000000 105.919998
2018-11-02 207.479996 1665.530029 150.350006 1057.790039 106.160004
2018-11-05 201.589996 1627.800049 148.679993 1040.089966 107.510002
# 直接用 matplotlib 畫它們的股價走勢
top_df['Close'].plot()
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x1a1741f748>
複製程式碼

Python 股票分析入門

# 根據他們的股價走勢,畫出股價波動
top_df_dr = top_df['Close'].pct_change()
top_df_dr.plot()
複製程式碼
<matplotlib.axes._subplots.AxesSubplot at 0x1a173004a8>
複製程式碼

Python 股票分析入門

# 上面的估計波動圖太密集,不容易開出問題,我們選亞馬遜和谷歌的進行對比
# 利用 seaborn 畫出亞馬遜和谷歌的散點圖
# 每個點對應的橫座標和縱座標,分別對應谷歌和亞馬遜當日的漲跌情況,如果都為負數說明當日股價均為下跌
sns.jointplot('AMZN', 'GOOG', top_df_dr)
複製程式碼
<seaborn.axisgrid.JointGrid at 0x1a172d8048>
複製程式碼

Python 股票分析入門

# 我們還可以利用強大的 seaborn 對五家公司進行相互對比
sns.pairplot(top_df_dr.dropna())
複製程式碼
<seaborn.axisgrid.PairGrid at 0x1a1786ada0>
複製程式碼

Python 股票分析入門

感謝閱讀!

相關文章