pytrends で Google トレンドの取得結果をマージする

% sw_vers
ProductName:    Mac OS X
ProductVersion: 10.13.6
BuildVersion:   17G65

# 初期化
from pytrends.request import TrendReq
from itertools import zip_longest
from itertools import filterfalse
import pandas as pd
pytrends = TrendReq(hl='ja-JP', tz=360)

# キーワードリスト
kw_list = ['Accelerated Mobile Pages','Apache HTTP Server','Args4j','Django','Dropwizard','EasyBatch','EclipseLink','Elasticsearch','FindBugs','Frontend QA(Quality Assurance)','Google Analytics','Google Apps Script','Google Data Studio','Google Search Console','Google Spread Sheet','Google Tag Manager','IntelliJ IDEA','Java','Jupyter Notebook','Mezzanine','Mixpanel','MongoDB','PWA','Page Speed Insights','Python','React','RedPen','SEO','Sonar','SonarQube','SonarQube squid','Wagtail CMS','Web API','Web Performance Test & Tips','Wicket','aXe','brew','gatsby','guice','jackson','linux','maven','memcached','msites','pmd','rundeck','security','sitespeed.io','spring-boot','wercker','データ分析','JavaScript','書籍','統計検定','資格取得','雑記']
# 主キーワード      
primary_kw = 'postgresql'
group_by = 4
chunks = zip_longest(*[iter(kw_list)]*group_by)
p = lambda x: x is None

merged_df = None
for elems in list(chunks):
    elems = list(filterfalse(p, elems))
    elems.append(primary_kw)
    pytrends.build_payload(elems, cat=0, timeframe='today 5-y', geo='JP', gprop='')
    df = pytrends.interest_over_time()
    # 取得結果のスコアは String になる。 float 変換したいので、True/False が設定されている `isPartial` を削除して float に変換する。   
    del df['isPartial']
    df = df.astype('float64')
    # dataframe　を primary_kw で最大値で正規化する 
    scaled_dataframe = df.div(df[primary_kw].max(), axis=0)
    if merged_df is None:
        merged_df = scaled_dataframe
    else:
        # ValueError: columns overlap but no suffix specified が発生するので、やむなく、'postgresql' を削除
        del scaled_dataframe[primary_kw]
        merged_df = merged_df.join(scaled_dataframe)

# Dataframe を表示
merged_df

%matplotlib inline
import matplotlib as mpl
mpl.rcParams['font.family'] = 'IPAPGothic'
# 数が多すぎるので出力対象を絞りこむ
keyword_list = ['Java','Apache HTTP Server','Dropwizard','Google Analytics','Mezzanine','postgresql','React','JavaScript','Python']
#keyword_list = ['Mezzanine','postgresql','React','JavaScript','Python']
merged_df[keyword_list].plot(logy=True, figsize=(15, 6))

<matplotlib.axes._subplots.AxesSubplot at 0x112642898>

merged_df.to_csv("google_trend.csv")