covid-19_sentiment.py

#Sorting And Feature Engineering
f_data = f_data.sort_values(by='date')
ft_data=f_data.copy()
ft_data['date'] = pd.to_datetime(f_data['date']).dt.date

ft_data['year']         = pd.DatetimeIndex(ft_data['date']).year
ft_data['month']        = pd.DatetimeIndex(ft_data['date']).month
ft_data['day']          = pd.DatetimeIndex(ft_data['date']).day
ft_data['day_of_year']  = pd.DatetimeIndex(ft_data['date']).dayofyear
ft_data['quarter']      = pd.DatetimeIndex(ft_data['date']).quarter
ft_data['season']       = ft_data.month%12 // 3 + 1

plt.subplot(2,1,1)
plt.title('Selecting A Cut-Off For Most Positive/Negative Tweets',fontsize=19,fontweight='bold')

ax0 = sns.kdeplot(f_data['Negative Sentiment'],bw=0.1)

kde_x, kde_y = ax0.lines[0].get_data()
ax0.fill_between(kde_x, kde_y, where=(kde_x>0.25) , 
                interpolate=True, color='b')

plt.annotate('Cut-Off For Most Negative Tweets', xy=(0.25, 0.5), xytext=(0.4, 2),
            arrowprops=dict(facecolor='red', shrink=0.05),fontsize=16,fontweight='bold')

ax0.axvline(f_data['Negative Sentiment'].mean(), color='r', linestyle='--')
ax0.axvline(f_data['Negative Sentiment'].median(), color='tab:orange', linestyle='-')
plt.legend({'PDF':f_data['Negative Sentiment'],r'Mean: {:.2f}'.format(f_data['Negative Sentiment'].mean()):f_data['Negative Sentiment'].mean(),
            r'Median: {:.2f}'.format(f_data['Negative Sentiment'].median()):f_data['Negative Sentiment'].median()})

plt.subplot(2,1,2)

ax1 = sns.kdeplot(f_data['Positive Sentiment'],bw=0.1,color='green')

plt.annotate('Cut-Off For Most Positive Tweets', xy=(0.4, 0.43), xytext=(0.4, 2),
            arrowprops=dict(facecolor='red', shrink=0.05),fontsize=16,fontweight='bold')
kde_x, kde_y = ax1.lines[0].get_data()
ax1.fill_between(kde_x, kde_y, where=(kde_x>0.4) , 
                interpolate=True, color='green')
ax1.set_xlabel('Sentiment Strength',fontsize=18)


ax1.axvline(f_data['Positive Sentiment'].mean(), color='r', linestyle='--')
ax1.axvline(f_data['Positive Sentiment'].median(), color='tab:orange', linestyle='-')
plt.legend({'PDF':f_data['Positive Sentiment'],r'Mean: {:.2f}'.format(f_data['Positive Sentiment'].mean()):f_data['Positive Sentiment'].mean(),
            r'Median: {:.2f}'.format(f_data['Positive Sentiment'].median()):f_data['Positive Sentiment'].median()})

plt.show()