Python：如何在一个直方图中绘制多个列

Question

我正在学习python，而我一直坚持的是基于FLT列绘制一个直方图来指示每个过滤器。我有这个数据框：

    VARLIST:      MJD FLT  FIELD  FLUXCAL  FLUXCALERR     MAG  MAGERR  \
0       OBS:  55161.6   g    NaN  -62.016      23.428     NaN  -0.410   
1       OBS:  55176.6   g    NaN   -8.183      21.252     NaN  -2.820   
2       OBS:  55179.6   g    NaN    0.451      19.109  -4.134  46.053   
3       OBS:  55188.6   g    NaN  511.964      21.218 -11.773   0.045   
4       OBS:  55206.6   g    NaN  682.704      22.329 -12.086   0.036   
..       ...      ...  ..    ...      ...         ...     ...     ...   
259     OBS:  56659.6   z    NaN  193.577      44.434 -10.717   0.249   
260     OBS:  56662.6   z    NaN    2.728      30.422  -6.089  12.109   
261     OBS:  56667.7   z    NaN   51.009      30.915  -9.269   0.658   
262     OBS:  56681.5   z    NaN    8.945      30.450  -7.379   3.696   
263     OBS:  56754.3   z    NaN   12.488      60.586  -7.741   5.268   

     peakMJD  mag_zpt27.5  snr_zpt27.5  magerr_zpt27.5  
0    55206.6    23.018741    -2.647089       -0.410262  
1    55206.6    25.217719    -0.385046       -2.820441  
2    55206.6    28.364559     0.023601       46.014133  
3    55206.6    20.726901    24.128759        0.045009  
4    55206.6    20.414419    30.574768        0.035519  
..       ...          ...          ...             ...  
259  55206.6    21.782866     4.356506        0.249282  
260  55206.6    26.410389     0.089672       12.110811  
261  55206.6    23.230883     1.649976        0.658191  
262  55206.6    25.121049     0.293760        3.696892  
263  55206.6    24.758768     0.206120        5.268770 

我的目标是绘制FLT列，但要指示不同的过滤器：g、i、r和z。通过搜索如何绘制直方图，我可以根据MAG（幅值）绘制基本的直方图。理想情况下，我想要的是4种不同的颜色来表示不同的过滤器。我知道该图将有重叠的磁过滤器，但我同意它，因为我需要一个视觉，我可以放大，如果我有必要。

下面的代码是我现在拥有的。现在这是最基本的，因为我一直在看留档，毫无进展。值得注意的是，中间有一堆关于我尝试的评论。一种是遍历每一行，查看过滤器，然后查看图表，但我不明白。另一个想法是创建4个新列，每个筛选器一个，然后绘制每个列。我被卡住了，想不通。

#HISTOGRAM
def plot_histogram(source, data_file):
    if source == 'villar':
        filename, ext = os.path.splitext(data_file)
        SnName_villar = filename[34:-6]

        # read in for clarifying info
        ps1_phot_info = pd.read_csv(data_file)

        ra_deg = ps1_phot_info.loc[2][0][11:19] #deg
        dec_deg = ps1_phot_info.loc[3][0][12:19] #deg
        final_z_villar = ps1_phot_info.loc[5][0][17:23] 

        # read in for data
        ps1_phot = pd.read_csv(data_file, skiprows=15, delim_whitespace=True)
        ps1_phot.drop(ps1_phot.tail(1).index,inplace=True)

        ps1_phot['peakMJD'] = ps1_phot.iloc[ps1_phot['FLUXCAL'].idxmax()]["MJD"] 
        
        # Calculate mag. Look at zeropoints (A. Villar = 32.5, YSE = 27.5)
        ps1_phot['mag_zpt27.5'] = np.array(-2.5*np.log10(np.abs(ps1_phot['FLUXCAL'])))+27.5
        ps1_phot['snr_zpt27.5'] = (ps1_phot['FLUXCAL'] / np.abs(ps1_phot['FLUXCALERR']))
        ps1_phot['magerr_zpt27.5'] = np.array(1.086 / ps1_phot['snr_zpt27.5'])



        mask = (ps1_phot['FLUXCAL'].notna()) #& (ps1_phot['FLUXCALERR'] <= 50) 
        ps1_masked = ps1_phot.loc[mask] #has mag obs and reasonable error

        print(ps1_masked)


        ps1_phot['peakMJD'] = ps1_phot.iloc[ps1_phot['FLUXCAL'].idxmax()]["MJD"] 
        print('here')
        print(ps1_phot)

        #Calculate mag. Look at zeropoints (A. Villar = 32.5, YSE = 27.5)
        ps1_phot['mag_zpt27.5'] = np.array(-2.5*np.log10(np.abs(ps1_phot['FLUXCAL'])))+27.5
        ps1_phot['snr_zpt27.5'] = (ps1_phot['FLUXCAL'] / np.abs(ps1_phot['FLUXCALERR']))
        ps1_phot['magerr_zpt27.5'] = np.array(1.086 / ps1_phot['snr_zpt27.5'])



        mask = (ps1_phot['FLUXCAL'].notna()) #& (ps1_phot['FLUXCALERR'] <= 50) 
        ps1_masked = ps1_phot.loc[mask] #has mag obs and reasonable error

        #print(ps1_masked)
        #ps1_phot['g_band'] = ps1_phot.iloc[ps1_phot['FLT']=='g']
        #print("THIS IS THE G BAND")
        #print(ps1_masked)
        #ps1_phot['r_band'] = ps1_phot.iloc[ps1_phot['FLT']=='r']
        #print("THIS IS THE R BAND")
        #print(ps1_masked)
        #ps1_phot['i_band'] = ps1_phot.iloc[ps1_phot['FLT']=='i']
        #print("THIS IS THE I BAND")
        #print(ps1_masked)
        #ps1_phot['z_band'] = ps1_phot.iloc[ps1_phot['FLT']=='z']
        #print("THIS IS THE Z BAND")
        #print(ps1_masked)
        
        #for pb in passbands:
        #    plt.hist(x=ps1_masked[''])

        #numpy.histogram(a, bins=10, range=None, normed=None, weights=None, density=None)[source]
        
        
        #passbands = ('g', 'r', 'i', 'z')
        #for pb in passbands:
        #    #x = passbands[pd]
            
        #    #blah blah code --> Plot histogram
        #    #errorbar only for scatter plot
        #    ax1.errorbar(x=ps1_masked[passbands[pd]]['MJD'] - ps1_masked[passbands[pd]]['peakMJD'], y=ps1_masked[passbands[pd]][f'{yaxis_is}'], yerr=ps1_masked[passbands[pd]][f'{yaxiserr_is}'],
        #    fmt='o', alpha=0.5, color='g', label=f'{x}-PS1')
        #ax1.ticklabel_format(useOffset=False, style='plain')
        
        yaxis_is = 'mag_zpt27.5' # or FLUXCAL
        yaxiserr_is = 'magerr_zpt27.5' # or FLUXCALERR
        
        fig = plt.figure(figsize=(12, 4))
        ax1 = fig.add_subplot(121)

        ax1.set_ylabel('Observations', fontsize=16)
        ax1.set_xlabel('Mag', fontsize=16)
        ax1.tick_params(labelsize=12)
        ax1.set_title(f'{SnName_yse}, z_yse={final_z_yse}; {SnName_villar}, z_villar={final_z_villar}', fontsize=16)

        #Plot

        ps1_phot['mag_zpt27.5'].hist()
        plt.savefig(f"path/to/file/my_plot_{pb}.png")

        plt.tight_layout()
        plt.show()

        fig.savefig(f'./Villar_Data_Graphs/{SnName_villar}_{SnName_yse}_{yaxis_is}_Histogram.png', format='png', bbox_inches='tight', dpi=300)    
        plt.close(fig) 
    
    
    #Read all the files "fileV = sorted(glob.glob('./Villar/ps1_sne_zenodo/*.dat'))" and then supernova type (has the same) 
    # check the redshift to see if the are comparable (range +/- 0.1)

这就是我调用上述函数的地方

x = './Villar/ps1_sne_zenodo/PS1_PS1MD_PSc000186.snana.dat'
SnName_villar, final_z_villar, ps1_masked_g, ps1_masked_r, ps1_masked_i, ps1_masked_z = get_dataframes(source='villar', data_file=x, plot_color='c', sntype_label='VIa')

y = './Photpipe/yselc_v3_photoz/GPC1v3_2020add.snana.dat'
SnName_yse, final_z_yse, pp_phot_g, pp_phot_r, pp_phot_i, pp_phot_z = get_dataframes(source='yse', data_file=y, plot_color='b', sntype_label='VIa')

plot_villar_and_yse(SnName_villar, final_z_villar, ps1_masked_g, ps1_masked_r, ps1_masked_i, ps1_masked_z, 
              SnName_yse, final_z_yse, pp_phot_g, pp_phot_r, pp_phot_i, pp_phot_z)


plot_histogram(source='villar', data_file=x)
#Call and plot histogram function

请让我知道你的想法和建议。

Answer 1

海运是一条路。要绘制发行版，您可以在海运库中使用几个选项：sns.distplot（）、sns.kdeplot（）、sns.boxplot（）、sns.histplot（）。就我个人而言，我喜欢使用sns.distplot（），因为它为您提供了显示直方图和曲线的选项。

Answer 2

这可以相对简单地使用Seborn的hist图函数来完成。

import pandas as pd
import numpy as np
import seaborn as sns

# construct some sample data
ps1_phot = pd.concat([pd.DataFrame({'FLT': flt, 'MAG': np.random.randn(1000) + i}) for i, flt in enumerate('girz')])

# plot overlapping histograms
sns.histplot(ps1_phot.dropna(subset=['MAG']), x='MAG', hue='FLT')

导致