Chart Type - Radial Bar Chart In Matplotlib (Python)
Here's a script that takes a data frame with two values, the current and benchmark, and returns radial bar charts to plot progress toward a goal. You can also choose a color using the color_theme parameter that takes values 'Grey', 'Purple, 'Blue', 'Green', 'Orange', or 'Red'. Reason to use it: One of the most common questions an analyst is asked is how close am I to a goal? We had a great thread about doing this in a variety of ways here. The most common tool that people default to on these is to use a gauge chart, which albeit feeling familiar with driving a car, is not necessarily appropriate for visually encoding in an elegant way. With Python in Sisense for Cloud Data Teams, we can start to create custom chart types that are awesome and fit use-cases perfectly. The biggest issue with gauge charts is when you go beyond 100%, do you loop back to the beginning of the gauge or just adjust where 100% is? If we take a step back and look at how the visual could work, we can assemble a few radial components into a full 360° = 100%. That way, when we want a 200%, we can just have multiple 360° rings. ####################################################################### ### LIBRARIES ### ####################################################################### import pandas as pd import matplotlib.pyplot as plt import numpy as np ####################################################################### ### HELPER FUNCTIONS ### ####################################################################### #USE: Create an array structure for rings. #INPUT: a df of row length 1 with the first column as the current metric value and the second colum is the target metric value #OUTPUT: an aray of arrays representing each ring def calculate_rings(df): if df.iloc[0,0] < df.iloc[0,1]: rings=[[df.iloc[0,0],df.iloc[0,1]-df.iloc[0,0]],[0,0]] elif df.iloc[0,0] / df.iloc[0,1] < 2: rings=[[df.iloc[0,0],0],[df.iloc[0,0] % df.iloc[0,1], df.iloc[0,1]-df.iloc[0,0] % df.iloc[0,1]]] else: rings = [[0,0],[0,0]] return rings #USE: Determine if the label for the rotating number label should be left/center/right #INPUT: a df of row length 1 with the first column as the current metric value and the second colum is the target metric value #OUTPUT: the proper text alignment def horizontal_aligner(df): metric = 1.0 * df.iloc[0,0] % df.iloc[0,1] / df.iloc[0,1] if metric in (0, 0.5): align = 'center' elif metric < 0.5: align = 'left' else: align = 'right' return align def vertical_aligner(df): metric = 1.0 * df.iloc[0,0] % df.iloc[0,1] / df.iloc[0,1] if metric < 0.25: align = 'bottom' elif metric < 0.75: align = 'top' elif metric > 0.75: align = 'bottom' else: align = 'center' return align #USE: Create a center label in the middle of the radial chart. #INPUT: a df of row length 1 with the first column as the current metric value and the second column is the target metric value #OUTPUT: the proper text label def add_center_label(df): percent = str(round(1.0*df.iloc[0, 0]/df.iloc[0, 1]*100,1)) + '%' return plt.text(0, 0.2, percent, horizontalalignment='center', verticalalignment='center', fontsize = 40, family = 'sans-serif') #USE: Formats a number with the apropiate currency tags. #INPUT: a currency number #OUTPUT: the properly formmated currency string def get_currency_label(num): currency = '' if num < 10**3: currency = '$' + str(num) elif num < 10**6: currency = '$' + str(round(1.0*num/10**3,1)) + 'K' elif df.iloc[0,0] < 10**9: currency = '$' + str(round(num/10**6,1)) + 'M' else: currency = '$' + str(round(num/10**9,1)) + 'B' return currency #USE: Create a dynamic outer label that servers a pointer on the ring. #INPUT: a df of row length 1 with the first column as the current metric value and the second column is the target metric value #OUTPUT: the proper text label at the apropiate position def add_current_label(df): currency = get_currency_label(df.iloc[0,0]) print('vertical: ' + vertical_aligner(df)) print('horizontal: ' + horizontal_aligner(df)) return plt.text(1.5 * np.cos(0.5 *np.pi - 2 * np.pi * (float(df.iloc[0,0]) % df.iloc[0,1] /df.iloc[0,1])), 1.5 * np.sin(0.5 *np.pi - 2 * np.pi * (float(df.iloc[0,0]) % df.iloc[0,1] / df.iloc[0,1])), currency, horizontalalignment=horizontal_aligner(df), verticalalignment=vertical_aligner(df), fontsize = 20, family = 'sans-serif') def add_sub_center_label(df): amount = 'Goal: ' + get_currency_label(df.iloc[0,1]) return plt.text(0, -.1, amount, horizontalalignment='center', verticalalignment='top', fontsize = 22,family = 'sans-serif') ####################################################################### ### MAIN FUNCTION ### ####################################################################### def create_radial_chart(df, color_theme = 'Purple'): # base styling logic color = plt.get_cmap(color_theme + 's') ring_width = 0.3 outer_radius = 1.5 inner_radius = outer_radius - ring_width # set up plot ring_arrays = calculate_rings(df) fig, ax = plt.subplots() if df.iloc[0, 0] > df.iloc[0, 1]: ring_to_label = 0 outer_edge_color = None inner_edge_color = 'white' else: ring_to_label = 1 outer_edge_color, inner_edge_color = ['white', None] # plot logic outer_ring, _ = ax.pie(ring_arrays[0],radius=1.5, colors=[color(0.9), color(0.15)], startangle = 90, counterclock = False) plt.setp( outer_ring, width=ring_width, edgecolor=outer_edge_color) inner_ring, _ = ax.pie(ring_arrays[1], radius=inner_radius, colors=[color(0.55), color(0.05)], startangle = 90, counterclock = False) plt.setp(inner_ring, width=ring_width, edgecolor=inner_edge_color) # add labels and format plots add_center_label(df) add_current_label(df) add_sub_center_label(df) ax.axis('equal') plt.margins(0,0) plt.autoscale('enable') return plt # call the chart maker function and display the chart periscope.output(create_radial_chart(df, color_theme='Purple')) # Currently supported color themes: Grey, Purple, Blue, Green, Orange, Red An additional layer of information can be introduced using a subtitle below the percentage. Thanks to Kyle Dempsey for developing a simple function to do this. Here's the code with an adjusted position of the percent and subtitle. def add_center_label(df): percent = str(round(1.0*df.iloc[0, 0]/df.iloc[0, 1]*100)) + '%' return plt.text(0, 0.2, percent, horizontalalignment='center', verticalalignment='center', fontsize = 40, family = 'sans-serif') def add_sub_center_label(df): amount = 'Goal: ' + get_currency_label(df.iloc[0,1]) return plt.text(0, -0.1, amount, horizontalalignment='center', verticalalignment='top', fontsize = 22,family = 'sans-serif')Funnel Charts In Python
Funnel charts are a great way to represent any drop-offs in sample size throughout a series of steps. Using a little bit of Python handiwork in Sisense for Cloud Data Teams' R/Python integration, we can easily create this chart type. A common use case for this chart type is to visualize your pipeline via a sales funnel. This is an extremely effective way to spot opportunities to improve your current sales process! Below is the Python snippet used to create the funnel chart above and the input parameters. Inputs df: a data frame from your SQL output with 2 columns: "step" (the parts of the process you want to visualize in a funnel chart) and "val" (the value associated with the Step in the funnel) show_n (optional parameter): boolean value that determines whether the user wants to display the value corresponding to each step. Default is set to True show_pct (optional parameter): parameter that determines whether the user wants to show each step as... A percent of the previous step ('of last step') A percent of the initial sample ('of whole') Show no percent values ('NA') - default value Snippet # SQL output is imported as a pandas dataframe variable called "df" import pandas as pd import matplotlib.pyplot as plt # Function: funnel_chart, creates a funnel chart image from a dataframe of steps # Inputs: (1) dataframe with 2 columns. The first is the "step", and the second is the "val" corresponding to each step. (2) Optional boolean parameter show_n (displays the n size of each step). (3) Optional paramater show_pct which allows the user to show each step of the funnel as a percent of the original n size ('of whole') or a percent or the previous step ('of last step') # Output: matplotlib image representing a funnel chart def funnel_chart(df, show_n=True,show_pct='NA'): #set up data frame df['val']=df['val'].astype(int) my_max=df['val'][0] df = df.sort_values('val',ascending=False).reset_index(drop=True) df['pct_of_whole']=round((df['val']/my_max)*100).astype(int) pct_change=[100] for j in range(1,len(df['val'])): pct_change.append(int(round(df['val'][j]/df['val'][j-1]*100))) df['pct_change']=pct_change df = df.sort_values('val').reset_index(drop=True) df['left']=(my_max-df['val'])/2 df['col']=['indigo','purple','darkviolet','DarkOrchid','MediumOrchid','orchid','plum','thistle'] #initialize plot fig, ax = plt.subplots() for i in range(len(df['step'])): ax.barh(0.5*i+0.5, df['val'][i], height=0.3, left=df['left'][i], align='center', color=df['col'][i],alpha = 1.0, label=df['step'][i]) if(show_n==True): ax.annotate(' ' + df['step'][i] + ': ' + str(df['val'][i]),xy=(my_max,0.5*i+0.45),horizontalalignment='left') else: ax.annotate(' ' + df['step'][i],xy=(my_max,0.5*i+0.45),horizontalalignment='left') if(show_pct=='of whole'): ax.annotate(str(df['pct_of_whole'][i]) + '%' ,xy=(my_max/2,0.5*i+0.45),horizontalalignment='center',color='white') elif(show_pct=='of last step'): ax.annotate(str(df['pct_change'][i]) + '%' ,xy=(my_max/2,0.5*i+0.45),horizontalalignment='center',color='white') #remove border and align chart ax.axis('off') fig.subplots_adjust(right=0.8) return fig # Use Sisense for Cloud Data Teams to visualize a dataframe or an image by passing data to periscope.output() periscope.output(funnel_chart(df,show_n=True,show_pct='of last step')) What other chart types do you like to use for your sales processes?
