통합데이터지도 로고

# 설치가 필요하다면 아래 문장으로 라이브러리 설치
# !pip pandas numpy matplotlib wordcloud bar_chart_race

import pandas as pd
import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt
import seaborn as sns
import wordcloud
import bar_chart_race as bcr
import warnings
warnings.filterwarnings("ignore")

import matplotlib.font_manager as fm

font_list = fm.findSystemFonts(fontpaths = None, fontext = 'ttf')

font_list
# print([i for i in font_list if 'kr' in i])

font_location = '/Users/subinium/Library/Fonts/NotoSansCJKkr-Light.otf'
font_name = fm.FontProperties(fname=font_location).get_name()
# print(f'시각화에 사용한 폰트는 {font_name} 입니다.')
mpl.rc('font', family=font_name)

# 시각화를 위한 설정
pd.options.display.max_columns = 999
mpl.rcParams['figure.dpi'] = 300
mpl.rcParams['axes.spines.top'] = False
mpl.rcParams['axes.spines.right'] = False

data = pd.read_csv('./ky_yngbgs_worry_rank_info_202012.csv')
data.head()

# 데이터의 전체 기간
st, ed = data['base_ym'].min(), data['base_ym'].max()
print(f'{st}부터 {ed}까지 수집한 데이터입니다.')

# 고민 전체 조회수
total_views = data['views_co'].sum()
print(f'총 조회수는 {total_views}회입니다.')

# 고민 키워드 종류와 개수
total_keywords = data['srchwrd_nm'].nunique()
print(f'청소년이 검색하는 고민 키워드는 총 {total_keywords}개입니다.')

from PIL import Image
from wordcloud import WordCloud, STOPWORDS

# 전체 텍스트 데이터 전처리
total_text = (data['srchwrd_nm'] + ' ') * data['views_co']
total_text = ' '.join(total_text)

# 워드 클라우드 시각화
fig, ax = plt.subplots(figsize=(12, 8))
wordcloud = WordCloud(width = 1000, height = 600, random_state=1,
                     background_color='black',
                     colormap=mpl.cm.twilight,
                     collocations=False,
                     font_path=font_location,
                     stopwords = STOPWORDS).generate(total_text)
ax.imshow(wordcloud)
ax.axis('off')
plt.show()

fig, ax = plt.subplots(1, 1, figsize=(17, 8))

kwd_cnt = data.groupby('srchwrd_nm')['views_co'].sum().sort_values(ascending=False)[:60]
color_tot = sns.color_palette("twilight_shifted", 60)
kwd_color = {k:c for k, c in zip(kwd_cnt.index, color_tot)}


kwd_cnt = kwd_cnt[:20]
ax.bar(kwd_cnt.index, kwd_cnt,
       color=sns.color_palette("twilight_shifted", 60)[:20],
       width=0.55,
       edgecolor='black',
       linewidth=0.7)

ax.margins(0.02, 0.05)

for i in range(20):
    ax.annotate(f'{kwd_cnt[i]}₩n({kwd_cnt[i]/kwd_cnt.sum()*100:.3}%)', xy=(i, kwd_cnt[i]+70),
                   va='center', ha='center',
                                  )

ax.set_ylim(0, 1790)
ax.set_title('청소년 고민 상위 20', weight='bold', fontsize=25, fontweight='bold')
ax.grid(axis='y', linestyle='-', alpha=0.4)

ax.set_ylabel('전체 조회수', fontsize=15)
ax.set_xlabel('키워드', fontsize=15)

fig.tight_layout()
plt.show()

# data_rank
# base_ym 데이터를 문자열에서 날짜포맷으로 변경
data['base_ym'] = pd.to_datetime(data['base_ym'], format='%Y%m')

# 막대그래프 레이스를 위한 전처리
data_rank = pd.pivot_table(data, columns='srchwrd_nm', index='base_ym', values='views_co').fillna(0)
data_rank_percentage = pd.pivot_table(data, columns='srchwrd_nm', index='base_ym', values='pt_value').fillna(0)
bcr.bar_chart_race(df=data_rank,
                   n_bars=10,
                   period_length=800,
#                    dpi=200,
                   cmap='twilight',
                   filename=None)

month_rank = data.pivot(columns='search_rank_nm', index='base_ym', values='srchwrd_nm').iloc[:,0:20]

# 연속 구간 추출 함수
def continuous_seq(kwd):
    arr = np.array(np.where(month_rank.values==kwd))
    _, length = arr.shape
    ret, st = [], 0
    for i in range(1, length):
        if arr[0, i-1]+1 != arr[0, i]:
            ret.append(arr[:, st:i])
            st = i
    ret.append(arr[:, st:])
    return ret

# 키워드 별 rank plot
def highlight(kwd, color, ax):
    lsts = continuous_seq(kwd)
    for lst in lsts:
        ax.plot(lst[0,:], lst[1,:],
                marker='o', ms=15, lw=2.5,
                color=color, zorder=0)

        ax.scatter(lst[0,:], lst[1,:],
                marker='o', s=110,
                color='white', zorder=10)
    return ax

# 전체 rank plot
def ranking(kwds):
    fig, axes = plt.subplots(2, 1, figsize=(30, 23))

    ax = axes[0]
    total_kwds = pd.Series(month_rank.values.flatten()).unique()

    for kwd in total_kwds:
        highlight(kwd, '#E0E0E0', ax)


    color_tot = sns.color_palette("twilight_shifted", 14)[::2]
    kwd_color = {k:c for k, c in zip(kwds, color_tot)}
    for kwd in kwds:
        highlight(kwd, kwd_color[kwd], ax)

    ax.invert_yaxis()
    ax.set_xticks(range(23))
    ax.set_xticklabels(pd.to_datetime(month_rank.index, format='%Y%m').strftime('%m\n%Y'), fontweight='bold', fontsize=15)
    ax.set_xlim(-0.2, 22.2)
    ax.set_yticks(range(20))
    ax.set_yticklabels([f'Rank {i}' for i in range(1, 21)], fontweight='bold', fontsize=15, color='gray')
    ax.spines['left'].set_visible(False)
    ax.spines['bottom'].set_visible(False)
    title_tmp = ['\"'+k+'\"' for k in kwds]
    ax.set_title(f'월별 {",".join(title_tmp)} 키워드 순위', weight='bold', fontsize=25, fontweight='bold')

    ax = axes[1]
    stack = pd.Series(np.ones(23))
    for kwd in kwds:
        b1 = ax.bar(range(23), data_rank_percentage[kwd], bottom=stack, color=kwd_color[kwd], width=0.6, label=kwd)
        ax.bar_label(b1, label_type='center', color='white', fontsize=13, fmt='%.1f%%')
        stack += data_rank_percentage[kwd].values
    ax.set_xticks(range(23))
    ax.set_xticklabels(pd.to_datetime(month_rank.index, format='%Y%m').strftime('%m\n%Y'), fontweight='bold', fontsize=15)
    ax.set_xlim(-0.4, 22.4)
    ax.set_yticks([])
    ax.legend(fontsize=15)
    ax.spines['left'].set_visible(False)
    ax.set_title(f'월별 {",".join(title_tmp)} 키워드 비율', weight='bold', fontsize=25, fontweight='bold')

    plt.show()

ranking(['친구'])

ranking(['공부', '시험', '학원'])

ranking(['전학','자퇴', '검정고시'])

ranking(['엄마', '아빠', '부모님'])

ranking(['임신'])

ranking(['생리', '초경'])

ranking(['주휴수당'])

ranking(['특성화고'])

ranking(['자해', '우울증'])

ranking(['숨쉬기', '스트레스'])