# python data seo analysis

Csv file
https://drive.google.com/open?id=181kBDudGYxBMEYUZT6BTv4hmMwNKtU63
我已經完成整個項目了，再請參考，如果有可以更改進的地方，希望可以分享給我~~
* def pivot_date_url =>完成Impressions, str to int，(read_csv後直接轉型態)
* def get_x_and_y => 轉備好 線性回歸的 datax, datay
* linear_regression_slope => 線性回歸，丟出斜率
型態的問題處理比較久

#!/usr/bin/env python
# coding: utf-8
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

get_ipython().run_line_magic('matplotlib', 'inline')
plt.rcParams['font.family']='DFKai-SB' #顯示中文 (for Win10)]

# https://vimsky.com/zh-tw/examples/detail/python-method-pandas.pivot_table.html
# https://www.codecademy.com/learn/data-processing-pandas/modules/dspath-intro-pandas
# gsc data source format
# column: Date(is Object;str), Page(is Object;str), Clicks(is Int), Impressions(is Object;str), CTR(is Object;str), Position(is float)

def pivot_date_url(file_path, value_matric='Impressions'): #<----匯入gsc檔案轉成，樞紐分析表，輸出檔案樞紐分析表
    gsc_df = pd.read_csv(file_path, encoding='utf8')
    print('各column type')
    print(gsc_df.dtypes)
    #Impression 將 str 1,423 => int 1432
    gsc_df['Impressions'] = gsc_df['Impressions'].str.replace(',','').astype(int)
    #開始轉置，樞紐 index = date
    gsc_df_date_url = gsc_df.pivot_table(values=value_matric, index='Date', columns='Page', aggfunc=sum)
    #replace NaN to 0
    gsc_df_date_url = gsc_df_date_url.fillna(0)
    page_list = list(gsc_df_date_url.columns)
    return gsc_df_date_url, list(page_list)


# codertw.com/程式語言/462517 有dataframe抓資料的方法，loc、iloc
def get_x_and_y(gsc_df_date_url, page='https://starthealthy.nestle.com.tw/',start_date='2019-10-01', end_date='2019-12-31'):
    # 強制 int化，datay 需要 int 型別，否則 np.array() 出錯
#     print(gsc_df_date_url.loc[start_date:end_date,page])
    datay = gsc_df_date_url.loc[start_date:end_date,page].astype(int)
#     把string中的千分位去除，如'2,019' =>'2019'，再astype(int)
#     datay = gsc_df_date_url.loc[start_date:end_date,page].astype('int')
#     https://ithelp.ithome.com.tw/articles/10194852
#     if isinstance(gsc_df_date_url.loc[end_date,page], str):
# #         print(type(gsc_df_date_url.loc[end_date,page]))
#         datay = gsc_df_date_url.loc[start_date:end_date,page].astype(int)
#         #檢查datay是否為str：是的話轉成int
#     else:
#         datay = gsc_df_date_url.loc[start_date:end_date,page]
#         #gsc是一個dataframe, 使用行、列的key搭配loc抓取特定範圍內的資料
#         #https://www.geeksforgeeks.org/python-check-if-a-variable-is-string/    
    start_day = 0
    end_day = 0
    x = 0
    for i in gsc_df_date_url.index:
        if i == start_date:
            start_day = x
        #print(start_day)
        elif i == end_date:
            end_day = x
        x+=1
        #print(end_day)
    datax = list(range(1,end_day-start_day+2))
    #https://www.codecademy.com/learn/learn-python-3/modules/learn-python3-lists
#     這裡的一樣輸出pandas dataframe
    return datax, datay

def linear_regression_slope (datax, datay, exponential=1):
#     np.array 放到這邊
    coefficients = np.polyfit(np.array(datax), np.array(datay), exponential) 
    p = np.poly1d(coefficients)
    return coefficients[0], coefficients, p


# In[2]:


def export_slope_as_df(file_path, value_matric='Impressions',start_date='2019-10-01', end_date='2019-12-31'):
    #read csv data export page_list, and timestamp data as gsc_df
    gsc_df, page_list = pivot_date_url(file_path)
    print(start_date + end_date)
    slopes = {"page":"slope"}
    for page in page_list:
        datax, datay = get_x_and_y(gsc_df, page)
        slope = linear_regression_slope(datax, datay)
        slopes.update({page:slope})
    return slopes


# In[3]:


nestle_slope_dict = export_slope_as_df("Nestle_181001_Raw_2.csv")


# In[4]:


print('列出所有資料，用{page:slope}的形式，dictionary')
print(nestle_slope_dict)


# In[5]:


print('列出以下網址的斜率，https://starthealthy.nestle.com.tw/section1.aspx')
print(nestle_slope_dict['https://starthealthy.nestle.com.tw/section1.aspx'][0])


# In[6]:

黃底就是斜率

if isinstance(gsc_df_date_url.loc[end_date,page], str):
        print(type(gsc_df_date_url.loc[end_date,page]))
        datay = gsc_df_date_url.loc[start_date:end_date,page].astype(int)
        #檢查datay是否為str：是的話轉成int
    else:
        datay = gsc_df_date_url.loc[start_date:end_date,page]
        #gsc是一個dataframe, 使用行、列的key搭配loc抓取特定範圍內的資料
        #https://www.geeksforgeeks.org/python-check-if-a-variable-is-string/    

可以考慮改成

# datay 需要 int 型別，否則 np.array() 出錯
datay = gsc_df_date_url.loc[start_date:end_date,page].astype(int)
# gsc是一個dataframe, 使用行、列的key搭配loc抓取特定範圍內的資料
# https://www.geeksforgeeks.org/python-check-if-a-variable-is-string/