Question

我正在编写一个用于游戏分析的简单工具，但是在Bokeh图中显示嵌套数据框时遇到了问题。当我选择基本类别参数时，例如“按游戏类型排序”

     total_players
type               
DM              184
TDM              52


<class 'pandas.core.frame.DataFrame'>

图形显示正确-https://imgur.com/kTSsvHl

但是当我尝试选择嵌套类别时，例如“按地图分类游戏类型”并按属性“和”计数

                   total_players
type map                        
DM   Cinema                  116
     Construction             14
     Lighthouse               10
     Ship                     44
TDM  Cinema                   38
     Lighthouse                8
     Ship                      6
<class 'pandas.core.frame.DataFrame'>

散景图将为空，无法显示

我想要得到这样的东西：https://imgur.com/dP7XBj5，但是使用'sum'方法进行汇总。资料集： Dataset

代码：

第一个“ if”块中的问题

def gamerounds_graph_create(category, attribute, sort_by):
current_feature_set = category.split("_") # transform from "cat1_cat2" label to cat list['cat1', 'cat2'] need for grouping
df = pd.read_sql(GameRounds.select().sql()[0], db.obj.connection()) # 0 - raw sql #TODO split data by chunks
df.to_csv('game_rounds.csv')
if attribute != 'total' and sort_by == '_count':
    """
    Problem in this 'if' block.
    """
    group = df.groupby(current_feature_set, as_index=True).agg({attribute: "sum"})

    source = ColumnDataSource().from_df(group) #Convert df to bokeh format
    ranges = df[current_feature_set[0]].unique()
    del df #Free RAM, df can be big


    plot = figure(plot_height=400, plot_width=1000, x_range=ranges,
              title="{attribute} by {category}".format(attribute=attribute, category=category))
    plot.vbar(source=source, x=category, top=attribute, width=0.8)

else:
    group = df.groupby(current_feature_set)
    factors = df[current_feature_set[0]].unique()
    print(type(group))
    del df #Free RAM, df can be big
    total_factors = len(factors)
    if total_factors < 3 or total_factors > 12:
        total_factors = 12

    palette = palette_dict['Set3'][total_factors]
    source = ColumnDataSource(group) #Convert df to bokeh format
    plot = figure(plot_height=400, plot_width=1000, x_range=group,
              title="{attribute} by {category}".format(attribute=attribute, category=category))

    factors_map = factor_cmap(category, factors=factors, palette=palette) #broken for nested categories
    if attribute == 'total':
        attribute = 'id' #Categories can be counted by numeric attribute only - id
        sort_by = '_count'
    plot.vbar(source=source, x=category, top=attribute.__add__(sort_by), width=0.8,
              fill_color = factors_map
              ) # inbuilt parameter of ColumnDataSource, already computed



plot.legend.orientation = 'horizontal'
return components(plot)

P.S如何为嵌套类别的vbar添加颜色？

如何显示按总和汇总的嵌套DataFrameGroupBy的bokeh vbar？

0 个答案: