按给定的间隔和计数对熊猫value_counts的数据进行量化、组合的绘图

Question

我试图使用Pandas value_counts作为输出，在python中创建一个条形图。对于背景数据，数据是来自森林火灾数据集的温度测量。下面是我用来获取bin_counts的代码(它是一个Series对象)

def discretizeData(cur_dataset, col_name, num_bins, bin_opts):
    '''
    cur_dataset: dataset containing values to be discretized
    col_name: column name for discretization
    bin_opts: array, containing either single value for number of bins, or 
    discretization_type: type of discretization (either 'equal-width' or 'equal-frequency')
    '''
    
    # Select specific data column for binning
    data_to_bin = cur_dataset[col_name]
    
    # Choose between equal width or equal frequency
    if bin_opts=='equal-frequency':
        # If "equal-frequency", we want to use the pandas qcut option for binning
        binned_data = pd.qcut(data_to_bin, q=num_bins)
    else:
        # Use equl-width instead
        binned_data = pd.cut(data_to_bin, bins=num_bins)
    
    # Get the bin counts to return; this will be more useful
    bin_counts = binned_data.value_counts().sort_index()
    
    return bin_counts

下面是一个使用等频率绑定选项运行该函数的示例：

fires = dataset_dict['forestfires']
col_name = 'temp'
num_bins = 5
bin_opts='equal-frequency'

输出是一个Pandas对象，其索引为Interval对象，并将该间隔计数为列值。看起来是这样的：

    Equal Frequency Binning Example:
(2.1990000000000003, 14.42]    104
(14.42, 17.9]                  104
(17.9, 20.6]                   106
(20.6, 23.58]                   99
(23.58, 33.3]                  104
Name: temp, dtype: int64

我还尝试将其转换为Pandas DataFrame，使用bin_width作为一列，count作为第二列，但我找不到任何能够处理间隔的绘图库。我已经试过了。有什么建议吗？

Answer 1

您需要手动绘制：

df = pd.DataFrame({'value': np.random.normal(10,size=1000)})

counts = discretizeData(df, 'value', 10, 'equal-frequency')

plt.bar([x.left for x in counts.index], counts, width=[x.right - x.left for x in counts.index])

输出：

​

​