python dataframe分箱操作一：cut分箱

1.等宽分箱

# ===========================方法一===============================
def binnings(data_res, b_for_filename):
    a = 'bins_by' # 分箱依据
    b = 'used_to_count' # 分箱之后用来统计数量的字段
    test_result_out = data_res[[b, a]]
    test_result_out['score'] = [int(i) for i in test_result_out[a]]
    # 分箱
    test_result_out['bin'] = pd.cut(list(test_result_out['score']), [i for i in range(300, 960, 30)])
    test_result_out['bin'] = test_result_out['bin'].astype(np.str)
    test_result_out['bin'] = test_result_out['bin'].replace('nan', -999) # 空值处理
    # 分箱之后的处理
    bins = test_result_out.groupby('bin', as_index=False).count()[['bin', b]]
    # bins[b].sum()
    bins['rate'] = bins[b]/bins[b].sum()
    bins.columns = ['bin', 'count', 'percentage']
    bins['percentage'] = [round(i, 3) for i in bins['percentage']] # 分箱之后的每个箱子所占百分比
    # 保存
    bins.to_excel('save_path/test_'+a+'_'+b_for_filename+'_bins.xlsx', index=False)

# ===========================方法二===============================
#获得分箱点
def get_break(data,num):
    
    score_cha = data.max()-data.min()
    ranges = math.floor(float(score_cha)/num)
    print(data.min(),data.max())
    index = []
    index.append(299)
    for i in range(int(data.min()+ranges),int(data.max()),ranges):
        index.append(i)
    index.append(900)

    print(index)
    return index

break_index = get_break(df[score],nbins)
df[new_score] = pd.cut(df[score],break_index,include_lowest=False, right=True)

# 可以通过labels参数来设置分箱之后的名称

In[43]: group_name = ['Youth', 'YoungAdult', 'MiddleAged', 'Senior']

In[45]: cats = pd.cut(ages, bins, labels=group_name)
In[47]: cats
Out[47]: 
[Youth, Youth, Youth, YoungAdult, Youth, ..., YoungAdult, Senior, MiddleAged, MiddleAged, YoungAdult]

Length: 12

Categories (4, object): [Youth < YoungAdult < MiddleAged < Senior]


In[46]: pd.value_counts(cats)
Out[46]: 
Youth         5
MiddleAged    3
YoungAdult    3
Senior        1
dtype: int64

2.等频分箱

data = np.random.randn(1000)
cats = pd.qcut(data, 4)
pd.value_counts(cats)