人口統計のcsvファイルをmatplotlibで描画→encoding='shift_jis'を指定しているのに日本語文字化け

Question

#やっていること
e-statより平成30年の各都道府県の人口（千人）をcsvファイルとしてダウンロードし、これをmatplotlibで描画しようとしています。Pythonを使って、まずdata frameとしてcsvファイルを読み込み、encoding='shift_jis'をここで指定しています。そして人口の数から,を抜き去り、いざ棒グラフにしようとすると、日本語部分が全て□□□と言った様に文字化けしてしまいます。

#該当コード・出力データ

"表章項目","男女別","人口","全国・都道府県","/時間軸（年）","平成22年","平成27年","平成28年","平成29年","平成30年"
"人口【千人】","男女計","総人口","全国","","128,057","127,095","126,933","126,706","126,443"
"人口【千人】","男女計","総人口","北海道","","5,506","5,382","5,352","5,320","5,286"
"人口【千人】","男女計","総人口","青森県","","1,373","1,308","1,293","1,278","1,263"
"人口【千人】","男女計","総人口","岩手県","","1,330","1,280","1,268","1,255","1,241"
"人口【千人】","男女計","総人口","宮城県","","2,348","2,334","2,330","2,323","2,316"
"人口【千人】","男女計","総人口","秋田県","","1,086","1,023","1,010","996","981"
"人口【千人】","男女計","総人口","山形県","","1,169","1,124","1,113","1,102","1,090"
"人口【千人】","男女計","総人口","福島県","","2,029","1,914","1,901","1,882","1,864"
"人口【千人】","男女計","総人口","茨城県","","2,970","2,917","2,905","2,892","2,877"
"人口【千人】","男女計","総人口","栃木県","","2,008","1,974","1,966","1,957","1,946"
"人口【千人】","男女計","総人口","群馬県","","2,008","1,973","1,967","1,960","1,952"
"人口【千人】","男女計","総人口","埼玉県","","7,195","7,267","7,289","7,310","7,330"
"人口【千人】","男女計","総人口","千葉県","","6,216","6,223","6,236","6,246","6,255"
"人口【千人】","男女計","総人口","東京都","","13,159","13,515","13,624","13,724","13,822"
"人口【千人】","男女計","総人口","神奈川県","","9,048","9,126","9,145","9,159","9,177"
"人口【千人】","男女計","総人口","新潟県","","2,374","2,304","2,286","2,267","2,246"
"人口【千人】","男女計","総人口","富山県","","1,093","1,066","1,061","1,056","1,050"
"人口【千人】","男女計","総人口","石川県","","1,170","1,154","1,151","1,147","1,143"
"人口【千人】","男女計","総人口","福井県","","806","787","782","779","774"
"人口【千人】","男女計","総人口","山梨県","","863","835","830","823","817"
"人口【千人】","男女計","総人口","長野県","","2,152","2,099","2,088","2,076","2,063"
"人口【千人】","男女計","総人口","岐阜県","","2,081","2,032","2,022","2,008","1,997"
"人口【千人】","男女計","総人口","静岡県","","3,765","3,700","3,688","3,675","3,659"
"人口【千人】","男女計","総人口","愛知県","","7,411","7,483","7,507","7,525","7,537"
"人口【千人】","男女計","総人口","三重県","","1,855","1,816","1,808","1,800","1,791"
"人口【千人】","男女計","総人口","滋賀県","","1,411","1,413","1,413","1,413","1,412"
"人口【千人】","男女計","総人口","京都府","","2,636","2,610","2,605","2,599","2,591"
"人口【千人】","男女計","総人口","大阪府","","8,865","8,839","8,833","8,823","8,813"
"人口【千人】","男女計","総人口","兵庫県","","5,588","5,535","5,520","5,503","5,484"
"人口【千人】","男女計","総人口","奈良県","","1,401","1,364","1,356","1,348","1,339"
"人口【千人】","男女計","総人口","和歌山県","","1,002","964","954","945","935"
"人口【千人】","男女計","総人口","鳥取県","","589","573","570","565","560"
"人口【千人】","男女計","総人口","島根県","","717","694","690","685","680"
"人口【千人】","男女計","総人口","岡山県","","1,945","1,922","1,915","1,907","1,898"
"人口【千人】","男女計","総人口","広島県","","2,861","2,844","2,837","2,829","2,817"
"人口【千人】","男女計","総人口","山口県","","1,451","1,405","1,394","1,383","1,370"
"人口【千人】","男女計","総人口","徳島県","","785","756","750","743","736"
"人口【千人】","男女計","総人口","香川県","","996","976","972","967","962"
"人口【千人】","男女計","総人口","愛媛県","","1,431","1,385","1,375","1,364","1,352"
"人口【千人】","男女計","総人口","高知県","","764","728","721","714","706"
"人口【千人】","男女計","総人口","福岡県","","5,072","5,102","5,104","5,107","5,107"
"人口【千人】","男女計","総人口","佐賀県","","850","833","828","824","819"
"人口【千人】","男女計","総人口","長崎県","","1,427","1,377","1,367","1,354","1,341"
"人口【千人】","男女計","総人口","熊本県","","1,817","1,786","1,774","1,765","1,757"
"人口【千人】","男女計","総人口","大分県","","1,197","1,166","1,160","1,152","1,144"
"人口【千人】","男女計","総人口","宮崎県","","1,135","1,104","1,096","1,089","1,081"
"人口【千人】","男女計","総人口","鹿児島県","","1,706","1,648","1,637","1,626","1,614"
"人口【千人】","男女計","総人口","沖縄県","","1,393","1,434","1,439","1,443","1,448"

import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_csv('./population.csv', encoding = 'shift_jis', index_col = '全国・都道府県')

df['平成30年'] = pd.to_numeric(df['平成30年'].str.replace(',', ''))

df['平成30年'].plot.bar()
plt.show()

#問題解決にあたり
Pythonファイルの別の箇所でencodingの引数をとる関数やモジュールがあるのかと自分なりに調べてみましたが、芳しい情報は得られませんでした。もしくは、それ以前の何か違う問題があるのでしょうか。素人質問で申し訳ございませんが、お力添え頂ける箇所がございましたら、ご教授のほどよろしくお願い申し上げます。

Answer 1

エンコーディングの問題でないのは、データを設定したデータフレームをprintなどで表示してみればわかります。

さて、matplotlibに対するフォントの設定ですが、正攻法で設定ファイルやフォントファイルを用意すると結構な手間がかかります。

それに比べると簡単なのは、japanize-matplotlibモジュールをインストールして、コードにimport japanize_matplotlibを追加することです。パッケージにはIPAexゴシックフォントも含まれています。

japanize-matplotlib · PyPI

これを利用するのであれば、pip install japanize-matplotlibとpip(Anacondaを使う人はconda)でインストールしておき、コードには、

import matplotlib.pyplot as plt
import japanize_matplotlib

のように、matplotlibの後でimportすれば、後はよろしくやってくれます。

Comments

[kay_ventris4]

japanize_matplotlibというモジュールは目にしていたのですが、そういうことだったんですね！日本語訳なんかいらんと勘違いして無視しておりました。ありがとうございました。

Answer 2

もしくは、それ以前の何か違う問題があるのでしょうか。

エンコーディングは関係ないです。
matplotlib のデフォルトフォントが英字フォントなので、日本語のテキストを描画したい場合は、フォントの設定が必要です。

matplotlib を日本語化する方法