Safe to Sleep via D'EVA Consulting

The goal of this project was to complete network analyses from an Excel data file with regards to NICHD and six parter organizations. This project was completed in three days under a narrow deadline. A lot of the code is rushed and unpolished, but ultimately provides an example of network analysis for real work in a short timeframe.

Steps for the project:

1. Wrote code to ingest and parse Excel file
   
2. Reviewed supporting material to understand the data, the organization, and the goals of the network analysis
   
3. Created visualizations for each of the categories provided in the Excel file
   
4. Added a write up that explained the approach, observations, and implications for future research.

!pip install xlrd

import matplotlib.pyplot as plt
import pandas as pd
import networkx as nx
import seaborn as sns
from networkx.algorithms.approximation import clique
from networkx.algorithms.clique import find_cliques
from networkx.algorithms.centrality import degree_centrality
from networkx.algorithms.cluster import clustering
from networkx.algorithms.cluster import average_clustering
pd.set_option('display.max_colwidth', -1)
pd.set_option('display.max_rows', None)

orgs = ['AAP','ACOG','CDC','CPSC','First Candle','HRSA','NICHD']
colors = sns.color_palette("hls", 7)
sns.palplot(sns.color_palette("hls", 7))

def lighten(colr, deg=.37):
    colr = tuple([c+deg for c in colr])
    return tuple([c if c<=1 else 1 for c in colr])

def darken(colr,deg=.37):
    colr = tuple([c-deg for c in colr])
    return tuple([c if c>=0 else 0 for c in colr])

sts_logo = '\n'+r'Safe to Sleep$^®$ Campaign:'+'\n'

def remove_str(s):
    if 'Safe Kids' in s:
        return 'Safe Kids'
    
    if 'Digital Media' in s:
        return 'Digital media'
    
    for o in orgs:
        if o in s:
            return o
    remove_l = ['• ','*']
    if 'Internal groups and committees focused on high-risk communities' in s:
        return 'Internal groups for high-risk'
    if '.org' in s:
        return 'Websites'
    if 'SUID/SIDS Project' in s:
        return 'SUID/SIDS Project'
    if 'National Fetal and Infant' in s:
        return 'NFIMR'
    if 'American Board of Obs' in s:
        return 'ABOG'
    if 'State and local' in s:
        return 'State/local health agencies'
    if 'National Center for Fatality' in s:
        return 'NCRFP'
    if 'Organizations that work with' in s:
        return 'Organizations that work with families'
    if 'State health offices and public health professionals' in s:
        return 'State health offices'
    if '(' in s:
        s = s.split(' (')[0]
    for r in remove_l:
        s = s.replace(r,'')
    if s[-1] == ' ':
        s = s[:-1]
    if s[0] == ' ':
        s = s[1:]
    if 'Grantees' in s:
        return 'Grantees'
    if s == 'Website' or s == 'website':
        return 'Websites'
    
    # techinical corrections
    if 'NAPPS-' in s:
        return 'NAPPSS-IIN'
    return s



def clean_data(cell):
    try:
        return [remove_str(i) for i in cell.split('\n') if len(i) >0 ]
    except AttributeError as e:
        return ['None']
    
excel_path = "net_an/NICHD_STS.xlsx"
sleep_df = pd.read_excel(excel_path)
sleep_df.columns = ['Category'] + [i for i in sleep_df.columns[1:]]
sleep_df = sleep_df.drop(sleep_df.index[-2:])
for c in sleep_df.columns:
    if c != 'Category':
        sleep_df[c] = sleep_df[c].apply(clean_data)

Adjacency Matrix

def create_adjacency_matrix(vals, filename):
    # create adjacency.txt for easy networkx import
    with open(f'{filename}.txt','w') as file:
        for l in vals:
            if len(l) > 1:
                file.write('|'.join(l)+'\n')
    
    return nx.read_adjlist(f'{file_name}.txt',delimiter='|')

def draw_all_types(G): draw_types = [nx.kamada_kawai_layout, nx.spring_layout, nx.shell_layout,nx.spectral_layout, nx.random_layout] for func in draw_types: # Begin Visualization fig = plt.figure(figsize=(15,15)) plt.title('Partners Adjacency',size=15,color='darkblue') # draw edges pos = func(G) nx.draw_networkx_edges(G,pos=pos, edge_color='gray') nx.draw_networkx_nodes(G,pos=pos,node_color='lightgreen',width=10) nx.draw_networkx_nodes(G,pos,nodelist=main_nodes,node_color='lightblue',width=10) nx.draw_networkx_nodes(G,pos,nodelist=investigate_nodes,node_color='yellow',width=10) nx.draw_networkx_nodes(G,pos,nodelist=orgs,node_color='blue',width=10) nx.draw_networkx_labels(G,pos=pos,font_size=10,font_color='black') plt.axis('off') plt.show()

Partners

Colors

Nodes

# create the adj_matrix list
col_org_list = list(zip(list(range(1,8)),orgs))
big_m =[]
for i,org in col_org_list:
    for val in sleep_df.iloc[5].values[i]:
        small_m = [org,val]
        big_m.append(small_m)

file_name ='partners_edgecount'                
G = create_adjacency_matrix(big_m,file_name)

G = nx.DiGraph(G)
main_nodes = [i for i in G.nodes if i in big_m[-1]]
main_nodes.append('NICHD')


node_count = {}
special_nodes = []
for val in list(G.nodes):
    n = len(G.edges(val))
    if n < 10 and n>1:
        special_nodes.append(val)
    if n == 1:
        n = ''
    node_count[val] = n

# Begin Visualization
fig = plt.figure(figsize=(30,28))
plt.title(f'{sts_logo}Organizations and their partners',size=15,color='black',fontsize=50)

pos = nx.kamada_kawai_layout(G)
pos['AAP'] = [pos['AAP'][0]-.2,pos['AAP'][1]]
pos['HRSA'] = [pos['HRSA'][0],pos['AAP'][1]-.25]



for i in range(1,8):
    edges=G.edges(orgs[i-1])
    nx.draw_networkx_edges(G,pos=pos,
                           style='dashed',
                           width=3,
                           arrow=False,
                           edgelist=edges,
                           edge_color=[colors[i-1] for n in edges], 
                           alpha=.8)

    

nx.draw_networkx_nodes(G,pos,nodelist=special_nodes,node_color='yellow',node_size=2200,alpha=1)
investigate_nodes = ['ASTHO','JPMA','NACCHO']
nx.draw_networkx_nodes(G,pos,nodelist=investigate_nodes,node_color='red',node_size=2200,alpha=1)  

for i in range(1,8):
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color='black',node_size=1500,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color='white',node_size=1300,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color=[colors[i-1]],node_size=1300,alpha=.5)

nx.draw_networkx_nodes(G,pos,nodelist=special_nodes,node_color='white',node_size=1500,alpha=.8)
    
for i in range(1,8):
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color='black',node_size=3500,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color=[colors[i-1]],node_size=3000,alpha=1)

    
nx.draw_networkx_labels(G,pos=pos,font_size=18,labels=node_count,font_weight='bold',font_color='black')

# clustering_labels = clustering(G)
# for k in list(clustering_labels.keys()):
#     clustering_labels[k] = round(clustering_labels[k], 3)
# nx.draw_networkx_labels(G,pos=pos,font_size=18,labels=clustering_labels,font_weight='bold',font_color='black')



for k in pos:
    pos[k] = [pos[k][0],pos[k][1]+.038]
    
for i in range(1,8):
    org_labels = {}
    for val in sleep_df.iloc[5].values[i]:
        org_labels[val] = val
    nx.draw_networkx_labels(G,
                            pos=pos,
                            font_size=18,
                            labels=org_labels,
                            font_color=darken(colors[i-1],.4),
                            bbox=dict(facecolor=lighten(colors[i-1],.4),
                                      alpha=.5,
                                      edgecolor=darken(colors[i-1],.5)))

for k in pos:
    pos[k] = [pos[k][0],pos[k][1]+.009]
    
for i in range(1,8):
    org_labels = {}
    org_labels[orgs[i-1]] = orgs[i-1]
    nx.draw_networkx_labels(G,pos=pos,font_size=35,labels=org_labels,font_color=darken(colors[i-1],.45),bbox=dict(facecolor=lighten(colors[i-1],.3),alpha=1,edgecolor=darken(colors[i-1],.5)))

plt.annotate(
s = f"\n Metric Used: Edge Count \n\n Average Clustering: {round(average_clustering(G),3)} \n Density: {round(nx.density(G),3)} \n", 
xy=(-.9, -.9), 
va='top',
ha='left',
fontsize = 20,
bbox=dict(facecolor='lightgray', alpha=.8,pad=.5,),
)       

plt.axis('off')
plt.tight_layout()
plt.savefig(f'final_visual/{file_name}.png',dpi=300)
plt.show()

Clustering Coefficient

# create the adj_matrix list
col_org_list = list(zip(list(range(1,8)),orgs))
big_m =[]
for i,org in col_org_list:
    for val in sleep_df.iloc[5].values[i]:
        small_m = [org,val]
        big_m.append(small_m)

file_name ='partners_clusteringcoef'                
G = create_adjacency_matrix(big_m,file_name)

G = nx.DiGraph(G)
main_nodes = [i for i in G.nodes if i in big_m[-1]]
main_nodes.append('NICHD')

node_count = {}
special_nodes = []
for val in list(G.nodes):
    n = len(G.edges(val))
    if n < 10 and n>1:
        special_nodes.append(val)
    if n == 1:
        n = ''
    node_count[val] = n

# Begin Visualization
fig = plt.figure(figsize=(30,28))
plt.title(f'{sts_logo}Organizations and their partners',size=15,color='black',fontsize=50)

pos = nx.kamada_kawai_layout(G)
pos['AAP'] = [pos['AAP'][0]-.2,pos['AAP'][1]]
pos['HRSA'] = [pos['HRSA'][0],pos['AAP'][1]-.25]



for i in range(1,8):
    edges=G.edges(orgs[i-1])
    nx.draw_networkx_edges(G,pos=pos,
                           style='dashed',
                           width=3,
                           arrow=False,
                           edgelist=edges,
                           edge_color=[colors[i-1] for n in edges], 
                           alpha=.8)

    

nx.draw_networkx_nodes(G,pos,nodelist=special_nodes,node_color='yellow',node_size=2200,alpha=1)
investigate_nodes = ['ASTHO','JPMA','NACCHO']
nx.draw_networkx_nodes(G,pos,nodelist=investigate_nodes,node_color='red',node_size=2200,alpha=1)  

for i in range(1,8):
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color='black',node_size=1500,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color='white',node_size=1300,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[5].values[i],node_color=[colors[i-1]],node_size=1300,alpha=.5)

nx.draw_networkx_nodes(G,pos,nodelist=special_nodes,node_color='white',node_size=1500,alpha=.8)
    
for i in range(1,8):
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color='black',node_size=3500,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color=[colors[i-1]],node_size=3000,alpha=1)

    
# nx.draw_networkx_labels(G,pos=pos,font_size=18,labels=node_count,font_weight='bold',font_color='black')

clustering_labels = clustering(G)
for k in list(clustering_labels.keys()):
    clustering_labels[k] = round(clustering_labels[k], 3)
nx.draw_networkx_labels(G,pos=pos,font_size=18,labels=clustering_labels,font_weight='bold',font_color='black')



for k in pos:
    pos[k] = [pos[k][0],pos[k][1]+.038]
    
for i in range(1,8):
    org_labels = {}
    for val in sleep_df.iloc[5].values[i]:
        org_labels[val] = val
    nx.draw_networkx_labels(G,
                            pos=pos,
                            font_size=18,
                            labels=org_labels,
                            font_color=darken(colors[i-1],.4),
                            bbox=dict(facecolor=lighten(colors[i-1],.4),
                                      alpha=.5,
                                      edgecolor=darken(colors[i-1],.5)))

for k in pos:
    pos[k] = [pos[k][0],pos[k][1]+.009]
    
for i in range(1,8):
    org_labels = {}
    org_labels[orgs[i-1]] = orgs[i-1]
    nx.draw_networkx_labels(G,pos=pos,font_size=35,labels=org_labels,font_color=darken(colors[i-1],.45),bbox=dict(facecolor=lighten(colors[i-1],.3),alpha=1,edgecolor=darken(colors[i-1],.5)))

plt.annotate(
s = f"\n Metric Used: Clustering Coefficient \n\n Average Clustering: {round(average_clustering(G),3)} \n Density: {round(nx.density(G),3)} \n", 
xy=(-.9, -.9), 
va='top',
ha='left',
fontsize = 20,
bbox=dict(facecolor='lightgray', alpha=.8,pad=.5,),
)   
    

plt.axis('off')
plt.tight_layout()
plt.savefig(f'final_visual/{file_name}.png')
plt.show()

Partners Informational

# create the adj_matrix list col_org_list = list(zip(list(range(1,8)),orgs)) big_m =[] for i,org in col_org_list: for val in sleep_df.iloc[5].values[i]: small_m = [org,val] big_m.append(small_m) file_name ='partners_relationship' G = create_adjacency_matrix(big_m,file_name) main_nodes = [i for i in G.nodes if i in big_m[-1]] main_nodes.append('NICHD') investigate_nodes = ['ASTHO','JPMA','NACCHO'] # Begin Visualization fig = plt.figure(figsize=(25,25)) plt.title('Organizations and Their Partners',size=15,color='black',fontsize=50) # draw edges pos = nx.spring_layout(G) nx.draw_networkx_edges(G,pos=pos, edge_color='gray') nx.draw_networkx_nodes(G,pos,nodelist=orgs,node_color='blue',node_size=1000) nx.draw_networkx_nodes(G,pos=pos,node_color='lightgreen',node_size=600) nx.draw_networkx_nodes(G,pos,nodelist=main_nodes,node_color='lightblue',node_size=600) nx.draw_networkx_nodes(G,pos,nodelist=investigate_nodes,node_color='yellow',node_size=800) for k in pos: pos[k] = [pos[k][0],pos[k][1]+.025] nx.draw_networkx_labels(G,pos=pos,font_size=12,font_color='black',bbox=dict(facecolor='white',alpha=.8,edgecolor='gray')) plt.axis('off') plt.savefig(f'final_visual/{file_name}.png') plt.show()

Targets

Merged

sts = [i for l in list(sleep_df.iloc[2].values[1:]) for i in l] primary = [i for l in list(sleep_df.iloc[0].values[1:]) for i in l] secondary = [i for l in list(sleep_df.iloc[1].values[1:]) for i in l] target_adjacency = [] full_list = list(set(primary+secondary)) make_full =[] for i in range(1,8): target_list = [] for j in [2,0,1]: target_list.append(list(sleep_df.iloc[j].values[i])) target_list_new = [] target_list_new.append([i for i in target_list[0]]) target_list_new.append([i for i in target_list[1]]) target_list_new.append(target_list[2]) make_full.append(target_list_new) get_uniques = [k for j in [j for i in make_full for j in i] for k in j] final_adjacency_matrix = [] for val in get_uniques: if val not in final_adjacency_matrix: final_adjacency_matrix.append(val) while final_adjacency_matrix: val = final_adjacency_matrix.pop() [target_adjacency.append([val,member]) for member in full_list] file_name ='target_visualize' G = create_adjacency_matrix(target_adjacency,file_name) pos = nx.circular_layout(G) fig = plt.figure(figsize=(15,15)) plt.suptitle(f"{sts_logo}Organizations and target audience",size=40,y=1.04) nx.draw_networkx_nodes(G,pos,node_color='gray',node_size=2600,alpha=.07,label='_nolegend_') for i in range(1,8): pos = nx.circular_layout(G) t_0 =make_full[i-1][0] t_1 =make_full[i-1][1] t_2 =make_full[i-1][2] labels = {} for val in t_0: labels[val] = val for val in t_1: labels[val] = val for val in t_2: labels[val] = val # draw edges nx.draw_networkx_nodes(G,pos,nodelist=t_0,node_color=[lighten(colors[i-1])],node_size=2500,alpha=.8) nx.draw_networkx_nodes(G,pos,nodelist=t_1,node_color=[(colors[i-1])],node_size=1200,alpha=1) nx.draw_networkx_nodes(G,pos,nodelist=t_2,node_color=[darken(colors[i-1])],node_size=500,alpha=1) for k in pos: pos[k] = [pos[k][0],pos[k][1]-.08] for i in range(1,8): org_labels = {} for l in make_full[i-1]: for val in l: org_labels[val] = val nx.draw_networkx_labels(G, pos=pos, font_size=9, labels=org_labels, font_color=darken(colors[i-1],.4), bbox=dict(facecolor=lighten(colors[i-1],.4), alpha=.5,edgecolor=darken(colors[i-1],.5))) plt.axis('off') # Plot legend. lgnd = plt.legend([' STS Targets',' Primary',' Secondary'], loc="center", frameon=False, labelspacing=4) for text in lgnd.get_texts(): text.set_fontsize(18) text.set_fontweight('bold') text.set_color(darken(colors[i-1],.4)) plt.tight_layout() plt.plot() plt.savefig(f'final_visual/{file_name}2.png') plt.show()

Individual

sts = [i for l in list(sleep_df.iloc[2].values[1:]) for i in l] primary = [i for l in list(sleep_df.iloc[0].values[1:]) for i in l] secondary = [i for l in list(sleep_df.iloc[1].values[1:]) for i in l] target_adjacency = [] full_list = list(set(primary+secondary)) make_full =[] for i in range(1,8): target_list = [] for j in [2,0,1]: target_list.append(list(sleep_df.iloc[j].values[i])) target_list_new = [] target_list_new.append([i for i in target_list[0]]) target_list_new.append([i for i in target_list[1]]) target_list_new.append(target_list[2]) make_full.append(target_list_new) get_uniques = [k for j in [j for i in make_full for j in i] for k in j] final_adjacency_matrix = [] for val in get_uniques: if val not in final_adjacency_matrix: final_adjacency_matrix.append(val) while final_adjacency_matrix: val = final_adjacency_matrix.pop() [target_adjacency.append([val,member]) for member in full_list] file_name ='target_visualize' G = create_adjacency_matrix(target_adjacency,file_name) pos = nx.circular_layout(G) fig = plt.figure(figsize=(22,45)) plt.suptitle(f"{sts_logo}Organizations and target audience",size=40,y=1.07) plt.tight_layout() for i in range(1,8): plt.subplot(4,2,i) plt.title(f'{orgs[i-1]} Targets',size=35,color=colors[i-1]) nx.draw_networkx_nodes(G,pos,node_color='gray',node_size=2600,alpha=.07,label='_nolegend_') t_0 =make_full[i-1][0] t_1 =make_full[i-1][1] t_2 =make_full[i-1][2] labels = {} for val in t_0: labels[val] = val for val in t_1: labels[val] = val for val in t_2: labels[val] = val # draw edges nx.draw_networkx_nodes(G,pos,nodelist=t_0,node_color=[lighten(colors[i-1])],node_size=2500,alpha=.8) nx.draw_networkx_nodes(G,pos,nodelist=t_1,node_color=[(colors[i-1])],node_size=1200,alpha=1) nx.draw_networkx_nodes(G,pos,nodelist=t_2,node_color=[darken(colors[i-1])],node_size=500,alpha=1) for k in pos: pos[k] = [pos[k][0],pos[k][1]-.08] nx.draw_networkx_labels(G,pos=pos,labels=labels, font_size=10, font_color=darken(colors[i-1],.5), font_weight='bold', bbox=dict(facecolor='white', alpha=.6, edgecolor='none')) plt.axis('off') # Plot legend. lgnd = plt.legend([' STS Targets',' Primary',' Secondary'], loc="center", frameon=False, labelspacing=4) for text in lgnd.get_texts(): text.set_fontweight text.set_fontweight('bold') text.set_color(darken(colors[i-1],.4)) plt.tight_layout() plt.plot() plt.savefig(f'final_visual/{file_name}.png') plt.show()

from matplotlib.patches import Patch
from matplotlib.lines import Line2D

columns = ['Audience','weight','Organization','audience']

all_targets = []

i=0
for l in list(sleep_df.iloc[2].values[1:]):  
    for m in l:
        all_targets.append([m,10, orgs[i],'STS'])
    i+=1
    
i=0
for l in list(sleep_df.iloc[0].values[1:]):  
    for m in l:
        all_targets.append([m,4, orgs[i], 'Primary'])
    i+=1
    
i=0
for l in list(sleep_df.iloc[1].values[1:]):  
    for m in l:
        all_targets.append([m,1, orgs[i],'Secondary'])
    i+=1


viz = pd.DataFrame(all_targets, columns=columns)
viz = viz[viz['Audience'] != 'None']

fig = plt.figure(figsize=(9,12))
fig.set_tight_layout(True)
plt.title(f"{sts_logo}Organizations and target audience",size=20,loc='center',x=0.2)
# plt.title(f'{check} Campaign:')
ax = plt.subplot(111)
g = sns.scatterplot(data=viz,
                x="Organization",
                y="Audience", 
                hue='Organization', 
                size="weight",
                style='audience',
                markers =['o','P','^'],
                sizes=(80, 325),
                alpha=.5,
                palette=colors)


legend_elements = [ Line2D([0], [0], marker='o', color='w', label='STS Target',
                          markerfacecolor='gray', markersize=8),
                   Line2D([0], [0], marker='P', color='w', label='Primary Target',
                          markerfacecolor='gray', markersize=10),
                    Line2D([0], [0], marker='^', color='w', label='Secondary Target',
                          markerfacecolor='gray', markersize=10)]

lgnd = ax.legend(handles=legend_elements, loc='lower right',title="Target Type:\n")


# lgnd.legendHandles[0].set_pos
    
# Hide the right and top spines
ax.spines['right'].set_visible(False)
ax.spines['top'].set_visible(False)

# Only show ticks on the left and bottom spines
ax.yaxis.set_ticks_position('left')
ax.xaxis.set_ticks_position('bottom')
plt.savefig(f'final_visual/target_scatter.png', dpi=300)
plt.show()

/Users/ryan/.local/share/virtualenvs/segmentation-APSAkCPI/lib/python3.7/site-packages/matplotlib/figure.py:98: MatplotlibDeprecationWarning: 
Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.
  "Adding an axes using the same arguments as a previous axes "
/Users/ryan/.local/share/virtualenvs/segmentation-APSAkCPI/lib/python3.7/site-packages/matplotlib/figure.py:2369: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.
  warnings.warn("This figure includes Axes that are not compatible "

sts = [i for l in list(sleep_df.iloc[2].values[1:]) for i in l] primary = [i for l in list(sleep_df.iloc[0].values[1:]) for i in l] secondary = [i for l in list(sleep_df.iloc[1].values[1:]) for i in l] target_adjacency = [] full_list = list(set(primary+secondary)) make_full =[] for i in range(1,8): target_list = [] for j in [2,0,1]: target_list.append(list(sleep_df.iloc[j].values[i])) target_list_new = [] target_list_new.append([i for i in target_list[0]]) target_list_new.append([i for i in target_list[1]]) target_list_new.append(target_list[2]) make_full.append(target_list_new) get_uniques = [k for j in [j for i in make_full for j in i] for k in j] final_adjacency_matrix = [] for val in get_uniques: if val not in final_adjacency_matrix: final_adjacency_matrix.append(val) while final_adjacency_matrix: val = final_adjacency_matrix.pop() [target_adjacency.append([val,member]) for member in full_list] file_name ='target_visualize' G = create_adjacency_matrix(target_adjacency,file_name) # pos = nx.circular_layout(G) fig = plt.figure(figsize=(20,40)) plt.suptitle(f"{sts_logo}Organizations and target audience",size=40,y=1.1) for i in range(1,8): plt.subplot(4,2,i) plt.tight_layout() plt.title(f'{orgs[i-1]} Targets',size=35,color=colors[i-1]) t_0 =make_full[i-1][0] t_1 =make_full[i-1][1] t_2 =make_full[i-1][2] labels = {} for val in t_0: labels[val] = val print(t_0) for val in t_1: labels[val] = val print(t_1) for val in t_2: print(t_2) labels[val] = val new_G = nx.Graph() [new_G.add_node(i) for i in t_0+t_1+t_2] pos_new = nx.circular_layout(new_G) # draw edges nx.draw_networkx_nodes(new_G,pos_new,nodelist=t_0,node_color=[lighten(colors[i-1])],node_size=2500,alpha=.8) nx.draw_networkx_nodes(new_G,pos_new,nodelist=t_1,node_color=[(colors[i-1])],node_size=1200,alpha=1) nx.draw_networkx_nodes(new_G,pos_new,nodelist=t_2,node_color=[darken(colors[i-1])],node_size=500,alpha=1) left_span = ['Pediatric care providers', 'Organizations that work with families', 'American Indians, and/or Alaska Natives', 'Technical assistance providers', 'People in residencies'] right_span = ['New and expectant parents', 'State health offices'] for k in t_0+t_1+t_2: if k in left_span: pos_new[k] = [pos_new[k][0]+.2,pos_new[k][1]] if k in right_span: pos_new[k] = [pos_new[k][0]-.2,pos_new[k][1]] nx.draw_networkx_labels(new_G,pos=pos_new,labels=labels, font_size=8, font_color=darken(colors[i-1],.5), font_weight='bold', bbox=dict(facecolor='white', alpha=.6, edgecolor='none')) plt.axis('off') # Plot legend. lgnd = plt.legend([' STS Targets',' Primary',' Secondary'], loc="center", frameon=False, labelspacing=4) for text in lgnd.get_texts(): text.set_fontweight text.set_fontweight('bold') text.set_color(darken(colors[i-1],.4)) plt.plot() plt.savefig(f'final_visual/{file_name}finalmerge.png') plt.show()

Outreach

# create the adj_matrix list
col_org_list = list(zip(list(range(1,8)),orgs))
big_m =[]
for i,org in col_org_list:
    for val in sleep_df.iloc[3].values[i]:
        small_m = [org,val]
        big_m.append(small_m)

file_name ='outreach_centrality'                
G = create_adjacency_matrix(big_m,file_name)

print(f"Average Clustering: {average_clustering(G)}")
print(f"Density: {nx.density(G)}")

# Begin Visualization
fig = plt.figure(figsize=(20,20))
plt.suptitle(f'{sts_logo}Outreach',size=15,color='black',fontsize=50)
plt.tight_layout()
# draw edges
pos = nx.kamada_kawai_layout(G,center=(0,0),scale=1.5)


# Customize positions for outlier nodes:
for v in sleep_df.iloc[3].values[6]:
    pos[v] = [pos[v][0],pos[v][1]-1.4]
pos[orgs[5]] = [pos[v][0]+.25,pos[v][1]+.2]

for v in sleep_df.iloc[3].values[7]:
    if v != 'Social media':
        pos[v] = [pos[v][0]-.35,pos[v][1]]
            
    if 'Information' in v or 'Partner' in v:
        pos[v] = [pos[v][0]-.12,pos[v][1]]

pos['Journals'] = [pos['Journals'][0],pos['Journals'][1]+.5]

for i in range(1,8):
    edges=G.edges(orgs[i-1])
    nx.draw_networkx_edges(G,pos=pos,
                           style='dashed',
                           width=3,
                           arrow=False,
                           edgelist=edges,
                           edge_color=[colors[i-1] for n in edges], 
                           alpha=.8)

for i in range(1,8):
#     nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[3].values[i],node_color='black',node_size=1500,alpha=1)
#     nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[3].values[i],node_color='white',node_size=1300,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=sleep_df.iloc[3].values[i],node_color=[colors[i-1]],node_size=1300,alpha=.5)

for i in range(1,8):
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color='black',node_size=2700,alpha=1)
    nx.draw_networkx_nodes(G,pos,nodelist=[orgs[i-1]],node_color=[colors[i-1]],node_size=2500,alpha=1)

node_count = {}
special_nodes = []
for val in list(G.nodes):
    n = len(G.edges(val))
    if n < 10 and n>1:
        special_nodes.append(val)
    if n == 1:
        n = ''
    node_count[val] = n
centrality_labels = degree_centrality(G)
for k in list(centrality_labels.keys()):
    centrality_labels[k] = round(centrality_labels[k], 2)
nx.draw_networkx_labels(G,pos=pos,font_size=10,labels=centrality_labels,font_weight='bold',font_color='black')

for k in pos:
    pos[k] = [pos[k][0],pos[k][1]-.09]
    
for i in range(1,8):
    org_labels = {}
    for val in sleep_df.iloc[3].values[i]:
        if val not in orgs:
            org_labels[val] = val
    nx.draw_networkx_labels(G,pos=pos,font_size=15,labels=org_labels,font_color=darken(colors[i-1],.4),bbox=dict(facecolor=lighten(colors[i-1],.4),alpha=.85,edgecolor=darken(colors[i-1],.5)))

for k in pos:
    pos[k] = [pos[k][0],pos[k][1]-.02]
    
for i in range(1,8):
    org_labels = {}
    org_labels[orgs[i-1]] = orgs[i-1]
    nx.draw_networkx_labels(G,pos=pos,font_size=19,labels=org_labels,font_color=darken(colors[i-1],.45),bbox=dict(facecolor=lighten(colors[i-1],.3),alpha=1,edgecolor=darken(colors[i-1],.5)))

plt.annotate(
s = f"\n Metric Used: Degree of Centrality \n\n Average Clustering: {round(average_clustering(G),3)} \n Density: {round(nx.density(G),3)} \n", 
xy=(-1.5, -1.5), 
va='top',
ha='left',
fontsize = 20,
bbox=dict(facecolor='lightgray', alpha=.8,pad=.5,),
)  
    
    
plt.axis('off')
plt.savefig(f'final_visual/{file_name}.png',dpi=300)
plt.show()

Average Clustering: 0.0
Density: 0.045068027210884355

/Users/ryan/.local/share/virtualenvs/segmentation-APSAkCPI/lib/python3.7/site-packages/networkx/drawing/nx_pylab.py:611: MatplotlibDeprecationWarning: isinstance(..., numbers.Number)
  if cb.is_numlike(alpha):

Assets

asset_lists = list(sleep_df.iloc[6].values[1:])
asset_lists_reduced = [['Education/Training',
  'Daily briefing with AAP members',
  'Hospital associations',
  'Mailing lists',
  'Publications',
  'Periodic survey',
  'Webpages'],
 ['Annual clinical meeting',
  'E-Learning',
  'Committee opinions and practice bulletins',
  'Publications',
  'Digital media',
  'Toolkits',
  'Webinars'],
 ['Clinician materials',
  'Grand Rounds',
  'Monthly presentations',
  'Publications',
  'Presentations for professional conferences',
  'SUID and SDY Case Registry/vital statistics',
  'Webpages'],
 ['Partner listserv',
  'Videos'],
 ['Education/Training',
  'Conferences',
  'Digital partnerships',
  'Educational materials',
  'Grand Rounds',
  'Guardian program',
  'PSAs',
  'Videos',
  'E-Learning',
  'Webinars'],
 ["Toolkits",
  'MCHB Challenges',
  "Educational materials"],
 ['Digital media',
  'Publications',
  'Toolkits',
  'Education/Training',
  'Videos',
  'Webinars']]
all_asset_lists = [i for j in asset_lists_reduced for i in j ]
assets_dict = {orgs[i]:asset_lists_reduced[i] for i in range(len(orgs))}
AG = nx.Graph(assets_dict)

plt.figure(figsize=(20,20))
plt.title(f'{sts_logo}Assets',size=15,color='black',fontsize=50)
plt.tight_layout()
post=nx.kamada_kawai_layout(AG)
AG = nx.Graph(assets_dict)
for i,org in enumerate(orgs):
    #add org nodes
    nx.draw_networkx_nodes(AG,post,
                   nodelist=[org],
                   node_color=[colors[i]],
                   node_size=2000,
               alpha=1)
    #add asset nodes
    nx.draw_networkx_nodes(AG,post,
                   nodelist=asset_lists_reduced[i],
                   node_color=[colors[i]],
                   node_size=1000,
               alpha=0.5)
    #add all edges
    edges=AG.edges([orgs[i]])
    nx.draw_networkx_edges(AG,pos=post,
                           style='dashed',
                           width=3,
                           arrow=False,
                           edgelist=edges,
                           edge_color=[colors[i] for n in edges], 
                           alpha=.8)
    #add additional features to asset nodes
    nx.draw_networkx_nodes(AG,post,
                           nodelist=asset_lists_reduced[i],
                           node_color='black',
                           node_size=1100,
                           alpha=1)
    nx.draw_networkx_nodes(AG,post,
                           nodelist=asset_lists_reduced[i],
                           node_color='white',node_size=1000,
                           alpha=1)
    nx.draw_networkx_nodes(AG,post,
                           nodelist=asset_lists_reduced[i],
                           node_color=[colors[i]],
                           node_size=1000,
                           alpha=.5)
    #add additional features to org nodes
    nx.draw_networkx_nodes(AG,post,
                           nodelist=[org],
                           node_color='black',
                           node_size=2100,alpha=1)
    nx.draw_networkx_nodes(AG,post,
                           nodelist=[org],
                           node_color=[colors[i]],
                           node_size=2000,
                           alpha=1)
node_count = {}
for val in list(AG.nodes):
    if len(AG.edges(val))>1:
        node_count[val] = len(AG.edges(val))
nx.draw_networkx_labels(AG,pos=post,
                        font_size=18,
                        labels=node_count,
                        font_color='black')


for k in post:
    post[k] = [post[k][0],post[k][1]+.05]
    
for i in range(1,8):
    org_labels = {val:val for val in asset_lists_reduced[i-1]}
    nx.draw_networkx_labels(AG,pos=post,
                            font_size=15,
                            labels=org_labels,
                            font_color=darken(colors[i-1],.4),
                            bbox=dict(facecolor=lighten(colors[i-1],.4),
                                      alpha=.85,edgecolor=darken(colors[i-1],.5)))
    
for i in range(1,8):
    org_labels = {}
    org_labels[orgs[i-1]] = orgs[i-1]
    nx.draw_networkx_labels(AG,pos=post,
                            font_size=19,
                            labels=org_labels,
                            font_color=darken(colors[i-1],.45),
                            bbox=dict(facecolor=lighten(colors[i-1],.3),
                                      alpha=1,edgecolor=darken(colors[i-1],.5)))

plt.annotate(
s = f"\n Metric Used: Edge Count \n\n Average Clustering: {round(average_clustering(G),3)} \n Density: {round(nx.density(G),3)} \n", 
xy=(.4, -.5), 
va='top',
ha='left',
fontsize = 20,
bbox=dict(facecolor='lightgray', alpha=.8,pad=.5,),
)      
    
plt.axis('Off')
plt.savefig('final_visual/asset_edgecount.png',dpi=300)