In [1]:
import pandas as pd
import janitor
import matplotlib.pyplot as plt
import seaborn as sns
from matplotlib.ticker import MaxNLocator
import math
import plotly.express as px
%matplotlib inline
In [2]:
sns.set_theme(context='notebook', style='ticks', palette='colorblind', font='sans-serif', font_scale=1, color_codes=True, rc=None)
sns.palplot(sns.color_palette())
In [3]:
outdir="wos_processed_data"

wos = pd.read_excel(f"../{outdir}/wos_processed.xlsx")
wos_univ = pd.read_excel(f"../{outdir}/wos_institution_locations_harmonized.xlsx")
In [4]:
wos_country = pd.read_excel(f"../{outdir}/wos_countries.xlsx")
wos_country_types = pd.read_excel(f"../{outdir}/wos_country_types.xlsx")
In [5]:
wos_country_types
Out[5]:
Country Country_Type
0 Belgium EU
1 China China
2 Luxembourg EU
3 Netherlands EU
4 Norway Non-EU associate
5 United Kingdom Non-EU associate
6 France EU
7 Sweden EU
8 Italy EU
9 Denmark EU
10 Germany EU
11 Slovenia EU
12 Estonia EU
13 Finland EU
14 Bulgaria EU
15 Slovakia EU
16 Spain EU
17 Poland EU
18 Czech Republic EU
19 Greece EU
20 Malta EU
21 Austria EU
22 Switzerland Non-EU associate
23 Ireland EU
24 Portugal EU
25 Romania EU
26 Hungary EU
27 Cyprus EU
28 Croatia EU
29 Lithuania EU
30 Latvia EU
In [6]:
# len(wos),len(wos_univ_locations)
In [7]:
# wos_addresses = pd.read_excel(f"/{outdir}/wos_addresses.xlsx")

# wos_affiliations = pd.read_excel(f"/{outdir}/wos_affiliations.xlsx")

# wos_author_locations = pd.read_excel(f"/{outdir}/wos_author_locations.xlsx")

# wos_univ_locations = pd.read_excel(f"/{outdir}/wos_univ_locations.xlsx")
In [8]:
record_col = "UT (Unique WOS ID)"

Output - per yer, by Metrix taxonomy¶

In [9]:
# def nth_repl_all(s, sub="", repl="<br>", nth=2):
#     find = s.find(sub)
#     # loop util we find no match
#     i = 1
#     while find != -1:
#         # if i  is equal to nth we found nth matches so replace
#         if i == nth:
#             s = s[:find]+repl+s[find + len(sub):]
#             i = 0
#         # find + len(sub) + 1 means we start after the last match
#         find = s.find(sub, find + len(sub) + 1)
#         i += 1
#     return s.replace("<br>&","&<br")

def replace_nth(s, sub=" ", repl="<br>", n=2):
    chunks = s.split(sub)
    size = len(chunks)
    rows = size // n + (0 if size % n == 0 else 1)
    return (repl.join([
        sub.join([chunks[i * n + j] for j in range(n if (i + 1) * n < size else size - i * n)])
        for i in range(rows)
    ])).replace("<br>&"," &<br>")


groups = ['Domain_English',"Field_English",'SubField_English']
data = wos.groupby(groups, as_index=False)[record_col].nunique().sort_values(ascending=False, by=record_col)
data["percent"] = data[record_col]/data[record_col].sum()

data[groups] = data[groups].applymap(replace_nth)
data
Out[9]:
Domain_English Field_English SubField_English UT (Unique WOS ID) percent
37 Applied Sciences Information &<br>Communication Technologies Artificial Intelligence &<br> Image<br>Processing 7915 0.171841
44 Applied Sciences Information &<br>Communication Technologies Networking &<br>Telecommunications 5360 0.116370
32 Applied Sciences Engineering Geological &<br>Geomatics Engineering 2576 0.055927
33 Applied Sciences Engineering Industrial Engineering &<br> Automation 2316 0.050282
15 Applied Sciences Enabling &<br>Strategic Technologies Energy 1965 0.042662
... ... ... ... ... ...
11 Applied Sciences Economics &<br>Business Business &<br>Management 1 0.000022
46 Applied Sciences Social Sciences Anthropology 1 0.000022
54 Arts &<br>Humanities Philosophy &<br>Theology Philosophy 1 0.000022
52 Arts &<br>Humanities Historical Studies History of<br>Social Sciences 1 0.000022
129 Health Sciences Psychology &<br>Cognitive Sciences General Psychology &<br> Cognitive<br>Sciences 1 0.000022

175 rows × 5 columns

In [10]:
fig = px.sunburst(data, path=groups, values=record_col,
                  color='Domain_English',title="Distribution of topics<br>(METRIX classification)", template='plotly')
fig.update_traces(hovertemplate='%{id}<br>%{value:d}')
fig.show(config= dict(displayModeBar = False))

Domains¶

In [96]:
group = 'Domain_English'
data = wos.groupby(group, as_index=False)[record_col].nunique().sort_values(ascending=False, by=record_col)
data
Out[96]:
Domain_English UT (Unique WOS ID)
0 Applied Sciences 29985
5 Natural Sciences 8457
3 Health Sciences 5341
2 Economic & Social Sciences 1360
4 Multidisciplinary 847
1 Arts & Humanities 70
In [ ]:
In [12]:
g = sns.barplot(data, x=record_col, y=group)
g.set_xlim(0,35000)
g.set_ylabel(None)
g.set_xlabel("Number of co-publications")
g.set_title("Distribution of Domains")
for i in g.containers:
    g.bar_label(i,fontsize=10)
In [13]:
fig = px.bar(data, x=record_col, y=group, color=group,
                              labels={
                     record_col: 'Number of co-publications',
                     group: "",
                 },
                title="Distribution of Domains", template='plotly')
fig.update_layout(showlegend=False, xaxis_tickformat='d',font_family="Montserrat")
fig.update_traces(hovertemplate='%{x:d}')
fig.add_shape(
        # Rectangle with reference to the plot
            type="rect",
            xref="paper",
            yref="paper",
            x0=0,
            y0=0,
            x1=1.0,
            y1=1.0,
            line=dict(
                color="black",
                 width=0.5,
             )
         )
fig.update_yaxes(
    showgrid=True,
    ticks="outside")
fig.update_xaxes(
    showgrid=True,
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [14]:
# # define a function to divide each row's 'Count' by the value of the first year
# def divide_by_first_year(group):
#     group['relative_growth'] = group[record_col] / group.loc[group['Publication Year'] == group['Publication Year'].min(), record_col].values[0]
#     return group
#
#
#
# data = (wos.groupby(group)[record_col].nunique()
#         .unstack(fill_value=0).stack()
#         .reset_index()
#         .rename(columns={0:record_col})
#         .sort_values(ascending=False, by=group+[record_col]))
#
# # group by 'Topic'
# grouped = data.groupby('Domain_English')
# # apply the function to each group
# data = grouped.apply(divide_by_first_year).reset_index(drop=True)
# data['relative_growth'] = data['relative_growth']*100
In [99]:
In [106]:
group = ['Publication Year','Domain_English']
data = (wos.groupby(['Publication Year','Domain_English'])[record_col].nunique(dropna=False).unstack()
        .fillna(0)
        .stack()
        .reset_index()
        .rename(columns={0:record_col}))
data = data.merge(data[data[record_col]>0].sort_values(by=["Publication Year"], ascending=True).drop_duplicates(subset='Domain_English'),
                  on='Domain_English', suffixes=[None,"_relative_growth"])
data[record_col+"_relative_growth"] = (data[record_col]-data[record_col+"_relative_growth"])/data[record_col+"_relative_growth"]*100
data
Out[106]:
Publication Year Domain_English UT (Unique WOS ID) Publication Year_relative_growth UT (Unique WOS ID)_relative_growth
0 2011 Applied Sciences 490.0 2011 0.000000
1 2012 Applied Sciences 593.0 2011 21.020408
2 2013 Applied Sciences 738.0 2011 50.612245
3 2014 Applied Sciences 1031.0 2011 110.408163
4 2015 Applied Sciences 1201.0 2011 145.102041
... ... ... ... ... ...
67 2018 Natural Sciences 753.0 2011 316.022099
68 2019 Natural Sciences 999.0 2011 451.933702
69 2020 Natural Sciences 1232.0 2011 580.662983
70 2021 Natural Sciences 1403.0 2011 675.138122
71 2022 Natural Sciences 1665.0 2011 819.889503

72 rows × 5 columns

In [107]:
g=sns.lineplot(data.sort_values(ascending=True, by=group[-1]),y=record_col,x=group[0], hue=group[-1], marker="o")
g.set(xticks=list(range(2012,2022+1,2)))
g.legend(title=None)
g.set_xlabel(None)
g.set_ylabel(None)
g.set_title("Yearly output of co-publications")
Out[107]:
Text(0.5, 1.0, 'Yearly output of co-publications')
In [108]:
fig = px.line(data.sort_values(ascending=[True,True], by=[group[0],group[-1]]),y=record_col,x=group[0], color=group[-1], markers=True,                             labels={
                     record_col: 'Number of co-publications',
                     group[-1]: "Domain",
                 },
                title="Yearly output of co-publications", template='plotly')
fig.update_traces(hovertemplate='%{y:d}')
fig.update_layout(hovermode='x unified')
fig.add_shape(
        # Rectangle with reference to the plot
            type="rect",
            xref="paper",
            yref="paper",
            x0=0,
            y0=0,
            x1=1.0,
            y1=1.0,
            line=dict(
                color="black",
                 width=0.5,
             )
         )
fig.update_yaxes(
    showgrid=True,
    ticks="outside")
fig.update_xaxes(
    showgrid=True,
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [109]:
fig = px.line(data.sort_values(ascending=[True,True], by=[group[0],group[-1]]),y=record_col+"_relative_growth",x=group[0], color=group[-1], markers=True,                             labels={
                     record_col+"_relative_growth": 'Rel. growth<br>in co-publications (%)',
                     group[-1]: "Domain",
                 },
                title="Relative growth in the output of co-publications", template='plotly')
fig.update_traces(hovertemplate='%{y:.2f}%')

fig.update_layout(hovermode='x unified',yaxis_tickformat='d',font_family="Montserrat")
fig.add_shape(
        # Rectangle with reference to the plot
            type="rect",
            xref="paper",
            yref="paper",
            x0=0,
            y0=0,
            x1=1.0,
            y1=1.0,
            line=dict(
                color="black",
                 width=0.5,
             )
         )
fig.update_yaxes(
    showgrid=True,
    ticks="outside")
fig.update_xaxes(
    showgrid=True,
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [17]:
In [18]:
pivot_data = pd.pivot_table(data, values=record_col, index=['Domain_English'],

                       columns=['Publication Year'], fill_value=0)
pivot_data
Out[18]:
Publication Year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Domain_English
Applied Sciences 490 593 738 1031 1201 1535 1920 2808 3729 4446 5295 6199
Arts & Humanities 0 0 0 4 1 3 7 4 11 11 16 13
Economic & Social Sciences 20 22 29 28 34 40 84 105 160 211 252 375
Health Sciences 116 120 155 184 216 243 321 403 611 755 1035 1182
Multidisciplinary 15 21 43 52 57 64 75 76 83 97 115 149
Natural Sciences 181 223 298 318 380 437 568 753 999 1232 1403 1665
In [19]:
f, ax = plt.subplots(figsize=(9, 6))
g = sns.heatmap(pivot_data, annot=True, fmt="d", linewidths=.5, ax=ax)
g.set(xlabel="", ylabel="")
Out[19]:
[Text(0.5, 33.249999999999986, ''), Text(79.74999999999999, 0.5, '')]
In [20]:
import numpy as np
percent_pivot = pd.crosstab(data['Domain_English'], data['Publication Year'], values=data[record_col], aggfunc=np.sum, normalize='columns')*100
percent_pivot
Out[20]:
Publication Year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Domain_English
Applied Sciences 59.610706 60.572012 58.432304 63.760049 63.578613 66.106804 64.537815 67.678959 66.672626 65.847156 65.241498 64.687467
Arts & Humanities 0.000000 0.000000 0.000000 0.247372 0.052938 0.129199 0.235294 0.096409 0.196674 0.162915 0.197141 0.135657
Economic & Social Sciences 2.433090 2.247191 2.296120 1.731602 1.799894 1.722653 2.823529 2.530730 2.860719 3.125000 3.104978 3.913180
Health Sciences 14.111922 12.257406 12.272367 11.379097 11.434621 10.465116 10.789916 9.713184 10.924370 11.181872 12.752587 12.334342
Multidisciplinary 1.824818 2.145046 3.404592 3.215832 3.017470 2.756245 2.521008 1.831767 1.483998 1.436611 1.416954 1.554837
Natural Sciences 22.019465 22.778345 23.594616 19.666048 20.116464 18.819983 19.092437 18.148952 17.861613 18.246445 17.286841 17.374517
In [21]:
f, ax = plt.subplots(figsize=(15, 6))
g = sns.heatmap(percent_pivot, annot=True, fmt='.2f', linewidths=.5, ax=ax, cbar=False)
for t in ax.texts: t.set_text(t.get_text() + " %")
g.set(xlabel="", ylabel="")
Out[21]:
[Text(0.5, 33.249999999999986, ''), Text(154.75, 0.5, '')]
In [22]:
percent_pivot.T.plot(kind='bar',
                    stacked=True,
                    figsize=(10, 6))
Out[22]:
<Axes: xlabel='Publication Year'>
In [23]:
percent_pivot.T.plot(kind='bar',
                        stacked=True,
                        figsize=(15, 8))

plt.legend(loc="lower left", ncol=2)
# plt.ylabel("Release Year")
# plt.xlabel("Proportion")


for n, x in enumerate([*pivot_data.T.index.values]):
    for (proportion, count, y_loc) in zip(percent_pivot.T.loc[x],
                                          pivot_data.T.loc[x],
                                          percent_pivot.T.loc[x].cumsum()):

        plt.text(y=(y_loc - proportion) + (proportion / 2),
                 x=n - 0.11,
                 s=f'{count}',# ({np.round(proportion, 1)}%)',
                 color="black",
                 fontsize=8,
                 fontweight="bold")

plt.show()

Field¶

In [24]:
group = ['Publication Year',"Domain_English",'Field_English']
data = wos.groupby(group, as_index=False)[record_col].nunique().sort_values(ascending=False, by=group+[record_col])
data
Out[24]:
Publication Year Domain_English Field_English UT (Unique WOS ID)
233 2022 Natural Sciences Physics & Astronomy 596
232 2022 Natural Sciences Mathematics & Statistics 228
231 2022 Natural Sciences Earth & Environmental Sciences 409
230 2022 Natural Sciences Chemistry 251
229 2022 Natural Sciences Biology 181
... ... ... ... ...
4 2011 Applied Sciences Information & Communication Technologies 256
3 2011 Applied Sciences Engineering 166
2 2011 Applied Sciences Enabling & Strategic Technologies 53
1 2011 Applied Sciences Built Environment & Design 6
0 2011 Applied Sciences Agriculture, Fisheries & Forestry 9

234 rows × 4 columns

In [25]:
len(data[group[-2]].unique())
Out[25]:
6
In [25]:
In [26]:
data_complete = pd.DataFrame()

for cat in sorted(data[group[-2]].unique()):
    #data segment
    sub_data = data[data[group[-2]]==cat]
    sub_data = sub_data.complete({group[0]:range(int(data[group[0]].min()), int(data[group[0]].max()) + 1)}
                                 ,group[-1],fill_value=0)
    data_complete = pd.concat([data_complete,sub_data], ignore_index=True)
    #plot
    g=sns.lineplot(sub_data.sort_values(ascending=True, by=group[-1]),
                   y=record_col,x=group[0], hue=group[-1], marker="o")
    g.set(xticks=list(range(2012,2022+1,2)))
    g.legend(title=None)
    g.set_title(cat)
    g.yaxis.set_major_locator(MaxNLocator(integer=True))
    plt.show()
In [27]:
data_complete = pd.DataFrame()

# Creating subplot axes
fig, axes = plt.subplots(nrows=3,ncols=2,figsize=(15, 15))

for cat,ax in zip(sorted(data[group[-2]].unique()),axes.flatten()):
    #data segment
    sub_data = data[data[group[-2]]==cat]
    sub_data = sub_data.complete({group[0]:range(int(data[group[0]].min()), int(data[group[0]].max()) + 1)}
                                 ,group[-1],fill_value=0)
    data_complete = pd.concat([data_complete,sub_data], ignore_index=True)
    #plot
    g=sns.lineplot(sub_data.sort_values(ascending=True, by=group[-1]),
                   y=record_col,x=group[0], hue=group[-1], marker="o", ax=ax)
    g.set(xticks=list(range(2012,2022+1,2)))
    g.legend(title=None)
    g.set_title(cat)
    g.set_xlabel(None)
    g.set_ylabel(None)
    g.yaxis.set_major_locator(MaxNLocator(integer=True))
fig.suptitle("Number of co-publications in domains and respective fields", y=0.92)
plt.show()

SubField¶

In [28]:
group = ['Publication Year',"Domain_English",'Field_English',"SubField_English"]
data = wos.groupby(group, as_index=False)[record_col].nunique().sort_values(ascending=False, by=group+[record_col])
data
Out[28]:
Publication Year Domain_English Field_English SubField_English UT (Unique WOS ID)
1598 2022 Natural Sciences Physics & Astronomy Optics 134
1597 2022 Natural Sciences Physics & Astronomy Nuclear & Particle Physics 65
1596 2022 Natural Sciences Physics & Astronomy Mathematical Physics 10
1595 2022 Natural Sciences Physics & Astronomy General Physics 31
1594 2022 Natural Sciences Physics & Astronomy Fluids & Plasmas 79
... ... ... ... ... ...
4 2011 Applied Sciences Agriculture, Fisheries & Forestry Forestry 1
3 2011 Applied Sciences Agriculture, Fisheries & Forestry Food Science 1
2 2011 Applied Sciences Agriculture, Fisheries & Forestry Fisheries 2
1 2011 Applied Sciences Agriculture, Fisheries & Forestry Dairy & Animal Science 2
0 2011 Applied Sciences Agriculture, Fisheries & Forestry Agronomy & Agriculture 3

1599 rows × 5 columns

In [29]:
for cat in sorted(data[group[-2]].unique()):
    sub_data = data[data[group[-2]]==cat]
    sub_data = sub_data.complete({group[0]:range(int(data[group[0]].min()), int(data[group[0]].max()) + 1)}
                                 ,group[-1],fill_value=0)
    g=sns.lineplot(sub_data.sort_values(ascending=True, by=group[-1]),y=record_col,x=group[0],
                   hue=group[-1], marker="o", errorbar=None)
    g.set(xticks=list(range(2012,2022+1,2)))
    g.legend(title=None,bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0., ncols=math.ceil(len(g.legend_.texts)/12))
    g.set_title(f'Number or co-publications in {cat}')
    g.set_ylabel(None)
    plt.show()
In [30]:
len(sorted(data[group[-2]].unique()))
Out[30]:
20
In [31]:
from  matplotlib.ticker import FuncFormatter
import math
def orderOfMagnitude(number):
    return math.floor(math.log(number, 10))

def roundToNearest(number):
    order = orderOfMagnitude(number)
    # if order!=0:
    #     order+=1
    near = math.ceil(number/10**order)*10**order
    return near

Collabs¶

In [32]:
wos_univ_locations = wos_univ.merge(wos_country_types, on="Country")
wos_univ_locations.sample(100)
Out[32]:
UT (Unique WOS ID) Institution Country Institution_harm merge_iter Country_Type
41191 WOS:000538161600016 Anhui Univ China Anhui Univ 0 China
175692 WOS:000709411500003 Univ Porto Portugal Univ Porto 0 EU
75198 WOS:000831217100027 Zhejiang Univ China Zhejiang Univ 0 China
48614 WOS:000597938400003 Shanghai Jiao Tong Univ China Shanghai Jiao Tong Univ 0 China
133670 WOS:000411824101159 Univ Pisa Italy Univ Pisa 0 EU
... ... ... ... ... ... ...
2892 WOS:000293708200019 Natl Univ Def Technol China Natl Univ Def Technol 0 China
125259 WOS:000663324800010 INRAE France INRAE 0 EU
55780 WOS:000659952900011 Huazhong Univ Sci & Technol China Huazhong Univ Sci & Technol 0 China
138600 WOS:000744399000001 Brignone Clin Italy Brignone Clin 0 EU
31040 WOS:000471758500010 Chinese Acad Sci China Chinese Acad Sci 0 China

100 rows × 6 columns

In [33]:
wos_collabs = wos_univ_locations[wos_univ_locations["Country_Type"]!="Other"][[record_col,"Country"]].drop_duplicates()
In [34]:
collab_desc = wos_collabs[wos_collabs["Country"]!="China"]["Country"].value_counts().reset_index()
collab_desc["percent_of_copubs"] = collab_desc["count"]/wos_collabs[record_col].nunique()*100
collab_desc["percent_contrib_in_copubs"] = collab_desc["count"]/wos_collabs[record_col].size*100
collab_desc = collab_desc.merge(wos_country_types, on="Country")
collab_desc

c_dict = {"count":"Number of co-publications",
          "percent_of_copubs":"Percent of co-publications",
          "percent_contrib_in_copubs":"Contribution to co-publications"}


# Creating subplot axes
# fig, axes = plt.subplots(ncols=3,figsize=(15, 15))
# for c,ax in zip(c_dict.keys(),axes.flatten()):
for c in c_dict.keys():
    data = collab_desc[["Country",c,"Country_Type"]]
    plt.figure(figsize=(9,12))
    g = sns.barplot(data, x=c, y="Country", hue="Country_Type", dodge=False)
    g.set_xlim(0,roundToNearest(data[c].max()))
    g.set_ylabel(None)
    g.set_xlabel(c_dict.get(c))
    g.set_title(c_dict.get(c))
    g.legend(title=None, loc="right")
    for i in g.containers:
        g.bar_label(i,fontsize=10, fmt='%.1f%%' if 'percent' in c else '%.0f')
    if 'percent' in c:
        g.xaxis.set_major_locator(MaxNLocator(integer=True))
        vals = g.get_xticks()
        g.set_xticklabels([str(int(val))+'%' for val in vals])
    plt.show()

C:\Users\radvanyi\AppData\Local\Temp\ipykernel_30956\556627507.py:29: UserWarning:

FixedFormatter should only be used together with FixedLocator


C:\Users\radvanyi\AppData\Local\Temp\ipykernel_30956\556627507.py:29: UserWarning:

FixedFormatter should only be used together with FixedLocator
In [35]:
wos_collabs_EU = wos_univ_locations[~wos_univ_locations["Country_Type"].isin(["Other","China"])][[record_col,"Country"]].drop_duplicates()
wos_collabs_EU = wos_collabs_EU.merge(wos_collabs_EU, on=record_col)
EU_co_occur = pd.crosstab(wos_collabs_EU['Country_x'], wos_collabs_EU['Country_y'], values=wos_collabs_EU[record_col], aggfunc='nunique', normalize='all').fillna(0)

# Generate a mask for the upper triangle
mask = np.triu(np.ones_like(EU_co_occur, dtype=bool))

# Set up the matplotlib figure
f, ax = plt.subplots(figsize=(11, 9))

# Draw the heatmap with the mask and correct aspect ratio
g = sns.heatmap(EU_co_occur, mask=mask,
            square=True, linewidths=.5)

g.set_ylabel(None)
g.set_xlabel(None)
Out[35]:
Text(0.5, 71.74999999999994, '')
In [36]:
wos_collabs_EU = wos_univ_locations[~wos_univ_locations["Country_Type"].isin(["Other","China"])][[record_col,"Country"]].drop_duplicates()
wos_collabs_EU = wos_collabs_EU.merge(wos_collabs_EU, on=record_col)
wos_collabs_EU
EU_co_occur = pd.crosstab(wos_collabs_EU['Country_x'], wos_collabs_EU['Country_y'], values=wos_collabs_EU[record_col], aggfunc='nunique').fillna(0).astype(int)


# Generate a mask for the upper triangle
mask = np.triu(np.ones_like(EU_co_occur, dtype=bool))
data = np.where(mask,None,EU_co_occur)
EU_co_occur.columns
Out[36]:
Index(['Austria', 'Belgium', 'Bulgaria', 'Croatia', 'Cyprus', 'Czech Republic',
       'Denmark', 'Estonia', 'Finland', 'France', 'Germany', 'Greece',
       'Hungary', 'Ireland', 'Italy', 'Latvia', 'Lithuania', 'Luxembourg',
       'Malta', 'Netherlands', 'Norway', 'Poland', 'Portugal', 'Romania',
       'Slovakia', 'Slovenia', 'Spain', 'Sweden', 'Switzerland',
       'United Kingdom'],
      dtype='object', name='Country_y')
In [37]:
fig = px.imshow(data,
                labels=dict(x="Country (x)", y="Country (y)", color="Co-publication"),
                x=list(EU_co_occur.columns),
                y=list(EU_co_occur.index), title="Intraeuropean patterns"
               )
fig.update_layout(title_x=0.5,
                   width=1000, height=1000,
                   xaxis_showgrid=False,
                   yaxis_showgrid=False,
                   yaxis_autorange='reversed', template='plotly_white')
fig.update_xaxes(tickangle= -90)
fig.update_yaxes(
    ticks="outside")
fig.update_xaxes(
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [38]:
collab_year = wos_collabs[wos_collabs["Country"]!="China"].copy()
collab_year = collab_year.merge(wos_country_types, on="Country").merge(wos[[record_col,"Publication Year"]],on=record_col).drop_duplicates()
data = collab_year.groupby(["Publication Year",'Country_Type'],as_index=False)[record_col].nunique()


g=sns.lineplot(data,y=record_col,x="Publication Year", hue="Country_Type", marker="o")
g.set(xticks=list(range(2012,2022+1,2)))
g.legend(title=None)
g.set_xlabel(None)
g.set_ylabel(None)
g.set_title("Yearly output of co-publications with China")
Out[38]:
Text(0.5, 1.0, 'Yearly output of co-publications with China')
In [87]:
import country_converter as coco
cc = coco.CountryConverter()

data = (collab_year.groupby(['Publication Year',"Country"])[record_col]
        .nunique(dropna=False).unstack()
        .fillna(0)
        .stack()
        .reset_index()
        .rename(columns={0:record_col}))
data = data.merge(data[data[record_col]>0].sort_values(by=["Publication Year"], ascending=True).drop_duplicates(subset="Country"),
                  on=["Country"], suffixes=[None,"_relative_growth"])
data[record_col+"_relative_growth"] = (data[record_col]-data[record_col+"_relative_growth"])/data[record_col+"_relative_growth"]*100
data
Out[87]:
Publication Year Country UT (Unique WOS ID) Publication Year_relative_growth UT (Unique WOS ID)_relative_growth
0 2011 Austria 22.0 2011 0.000000
1 2012 Austria 24.0 2011 9.090909
2 2013 Austria 26.0 2011 18.181818
3 2014 Austria 39.0 2011 77.272727
4 2015 Austria 50.0 2011 127.272727
... ... ... ... ... ...
355 2018 United Kingdom 1837.0 2011 406.060606
356 2019 United Kingdom 2430.0 2011 569.421488
357 2020 United Kingdom 3108.0 2011 756.198347
358 2021 United Kingdom 3718.0 2011 924.242424
359 2022 United Kingdom 4245.0 2011 1069.421488

360 rows × 5 columns

In [88]:
data["ISO3"] = cc.pandas_convert(series=data["Country"], to='ISO3')
fig = px.choropleth(data, locations="ISO3", color=record_col, hover_name="Country",
                    animation_frame='Publication Year', scope="europe", template='plotly', range_color=[data[record_col].min(),data[record_col].max()])
fig.show()
In [89]:
data["ISO3"] = cc.pandas_convert(series=data["Country"], to='ISO3')
fig = px.choropleth(data, locations="ISO3", color=record_col+"_relative_growth", hover_name="Country",
                    animation_frame='Publication Year', scope="europe", template='plotly',
                    range_color=[data[record_col+"_relative_growth"].min(),data[record_col+"_relative_growth"].max()])
fig.show()
In [90]:
[data[record_col+"_relative_growth"].min(),data[record_col+"_relative_growth"].max()]
Out[90]:
[-100.0, 3700.0]
In [91]:
fig = px.line(data.sort_values(ascending=True, by='Publication Year'),y=record_col,x='Publication Year', color="Country", markers=True,
              labels={
                     record_col: 'Number of co-publications',
                 },
                title="Yearly output of co-publications", template='plotly')
fig.update_traces(hovertemplate='%{y:d}')
fig.update_layout(hovermode='x unified')
fig.add_shape(
        # Rectangle with reference to the plot
            type="rect",
            xref="paper",
            yref="paper",
            x0=0,
            y0=0,
            x1=1.0,
            y1=1.0,
            line=dict(
                color="black",
                 width=0.5,
             )
         )
fig.update_yaxes(
    showgrid=True,
    ticks="outside")
fig.update_xaxes(
    showgrid=True,
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [92]:
fig = px.line(data.sort_values(ascending=True, by='Publication Year'),y=record_col+"_relative_growth",x='Publication Year', color="Country", markers=True,
              labels={
                     record_col+"_relative_growth": 'Relative growth of co-publications (%)',
                 },
                title="Relative growth of co-publications<br>(baseline: 2011)", template='plotly')
fig.update_traces(hovertemplate='%{y:d}%')
fig.add_shape(
        # Rectangle with reference to the plot
            type="rect",
            xref="paper",
            yref="paper",
            x0=0,
            y0=0,
            x1=1.0,
            y1=1.0,
            line=dict(
                color="black",
                 width=0.5,
             )
         )
fig.update_yaxes(
    showgrid=True,
    ticks="outside")
fig.update_xaxes(
    showgrid=True,
    ticks="outside")
fig.show(config= dict(displayModeBar = False))
In [40]:
year_pivot = pd.crosstab(collab_year['Country'], collab_year['Publication Year'], values=collab_year[record_col], aggfunc='nunique').fillna(0).astype(int)
year_pivot
Out[40]:
Publication Year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Country
Austria 22 24 26 39 50 57 72 89 138 137 185 205
Belgium 34 38 40 65 71 81 90 133 179 213 242 292
Bulgaria 4 5 8 9 7 19 21 18 10 25 32 19
Croatia 1 2 6 8 10 7 10 19 27 29 33 35
Cyprus 2 1 5 5 5 5 8 7 15 28 36 43
Czech Republic 13 15 16 21 20 36 37 56 64 81 93 123
Denmark 35 33 40 59 68 74 101 195 234 245 293 343
Estonia 3 3 7 10 12 10 15 15 16 38 45 39
Finland 31 35 44 82 100 125 126 198 241 256 289 380
France 117 130 174 231 269 325 348 491 648 691 807 858
Germany 123 172 192 273 310 365 456 604 801 907 1210 1386
Greece 15 18 19 32 35 50 47 81 114 122 139 181
Hungary 11 11 21 16 20 38 34 47 61 61 83 90
Ireland 13 16 22 31 27 45 66 72 84 116 167 187
Italy 51 70 84 116 178 187 247 325 441 571 641 811
Latvia 0 0 1 0 1 8 10 15 10 9 13 18
Lithuania 1 2 10 4 4 13 12 23 38 36 38 38
Luxembourg 2 3 3 1 8 9 13 15 18 22 35 51
Malta 1 0 0 0 1 1 0 0 6 2 7 10
Netherlands 72 64 77 103 139 166 220 297 408 470 529 655
Norway 30 42 60 76 67 88 104 134 222 253 304 311
Poland 17 31 37 57 73 82 98 110 138 181 276 353
Portugal 16 23 35 41 45 58 79 119 136 147 204 212
Romania 7 15 13 16 25 26 37 57 64 55 48 62
Slovakia 9 6 6 10 12 22 18 27 27 34 36 45
Slovenia 7 7 10 12 17 27 22 47 54 31 48 40
Spain 50 49 69 112 138 185 232 273 356 386 473 640
Sweden 34 50 59 83 113 170 233 232 385 359 428 510
Switzerland 37 50 54 74 74 95 155 195 233 263 349 447
United Kingdom 363 417 531 660 781 979 1350 1837 2430 3108 3718 4245
In [41]:
f, ax = plt.subplots(figsize=(15, 15))
g = sns.heatmap(year_pivot, annot=True, fmt="d", linewidths=.5, ax=ax)
g.set(xlabel="", ylabel="")
for i in range(year_pivot.shape[0]+1):
    ax.axhline(i, color='white', lw=10)
In [42]:
year_percent_pivot = pd.crosstab(collab_year['Country'], collab_year['Publication Year'], values=collab_year[record_col], aggfunc='nunique', normalize='columns').fillna(0)*100
year_percent_pivot
Out[42]:
Publication Year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Country
Austria 1.962533 1.801802 1.557819 1.736420 1.865672 1.699970 1.689744 1.552958 1.816267 1.543488 1.712804 1.623248
Belgium 3.033006 2.852853 2.396645 2.894034 2.649254 2.415747 2.112180 2.320712 2.355883 2.399730 2.240533 2.312139
Bulgaria 0.356824 0.375375 0.479329 0.400712 0.261194 0.566657 0.492842 0.314081 0.131614 0.281658 0.296269 0.150447
Croatia 0.089206 0.150150 0.359497 0.356189 0.373134 0.208768 0.234687 0.331530 0.355357 0.326724 0.305527 0.277140
Cyprus 0.178412 0.075075 0.299581 0.222618 0.186567 0.149120 0.187749 0.122143 0.197420 0.315457 0.333302 0.340486
Czech Republic 1.159679 1.126126 0.958658 0.934996 0.746269 1.073665 0.868341 0.977142 0.842327 0.912573 0.861031 0.973949
Denmark 3.122212 2.477477 2.396645 2.626892 2.537313 2.206979 2.370336 3.402548 3.079758 2.760252 2.712712 2.715971
Estonia 0.267618 0.225225 0.419413 0.445236 0.447761 0.298240 0.352030 0.261734 0.210582 0.428121 0.416628 0.308813
Finland 2.765388 2.627628 2.636309 3.650935 3.731343 3.728005 2.957052 3.454894 3.171887 2.884182 2.675678 3.008948
France 10.437110 9.759760 10.425404 10.284951 10.037313 9.692812 8.167097 8.567440 8.528560 7.785038 7.471530 6.793887
Germany 10.972346 12.912913 11.503895 12.154942 11.567164 10.885774 10.701713 10.539173 10.542248 10.218567 11.202666 10.974741
Greece 1.338091 1.351351 1.138406 1.424755 1.305970 1.491202 1.103027 1.413366 1.500395 1.374493 1.286918 1.433209
Hungary 0.981267 0.825826 1.258238 0.712378 0.746269 1.133313 0.797935 0.820101 0.802843 0.687247 0.768447 0.712645
Ireland 1.159679 1.201201 1.318155 1.380232 1.007463 1.342082 1.548932 1.256325 1.105554 1.306895 1.546153 1.480719
Italy 4.549509 5.255255 5.032954 5.164737 6.641791 5.577095 5.796761 5.670913 5.804159 6.433078 5.934636 6.421728
Latvia 0.000000 0.000000 0.059916 0.000000 0.037313 0.238592 0.234687 0.261734 0.131614 0.101397 0.120359 0.142529
Lithuania 0.089206 0.150150 0.599161 0.178094 0.149254 0.387712 0.281624 0.401326 0.500132 0.405588 0.351819 0.300895
Luxembourg 0.178412 0.225225 0.179748 0.044524 0.298507 0.268416 0.305093 0.261734 0.236904 0.247859 0.324044 0.403832
Malta 0.089206 0.000000 0.000000 0.000000 0.037313 0.029824 0.000000 0.000000 0.078968 0.022533 0.064809 0.079183
Netherlands 6.422837 4.804805 4.613541 4.585931 5.186567 4.950790 5.163107 5.182342 5.369834 5.295178 4.897695 5.186476
Norway 2.676182 3.153153 3.594967 3.383793 2.500000 2.624515 2.440742 2.338161 2.921822 2.850383 2.814554 2.462586
Poland 1.516503 2.327327 2.216896 2.537845 2.723881 2.445571 2.299930 1.919386 1.816267 2.039207 2.555319 2.795154
Portugal 1.427297 1.726727 2.097064 1.825467 1.679104 1.729794 1.854025 2.076426 1.789945 1.656151 1.888714 1.678676
Romania 0.624442 1.126126 0.778910 0.712378 0.932836 0.775425 0.868341 0.994591 0.842327 0.619648 0.444403 0.490934
Slovakia 0.802855 0.450450 0.359497 0.445236 0.447761 0.656129 0.422436 0.471122 0.355357 0.383055 0.333302 0.356323
Slovenia 0.624442 0.525526 0.599161 0.534283 0.634328 0.805249 0.516311 0.820101 0.710713 0.349256 0.444403 0.316731
Spain 4.460303 3.678679 4.134212 4.986643 5.149254 5.517447 5.444731 4.763567 4.685444 4.348806 4.379224 5.067701
Sweden 3.033006 3.753754 3.535051 3.695459 4.216418 5.070086 5.468200 4.048159 5.067123 4.044615 3.962596 4.038324
Switzerland 3.300624 3.753754 3.235470 3.294746 2.761194 2.833284 3.637644 3.402548 3.066596 2.963046 3.231182 3.539473
United Kingdom 32.381802 31.306306 31.815458 29.385574 29.141791 29.197733 31.682704 32.053743 31.982101 35.015773 34.422739 33.613113
In [43]:
f, ax = plt.subplots(figsize=(15, 15))
g = sns.heatmap(year_percent_pivot, annot=True, fmt='.1f', linewidths=(.5), ax=ax, cbar=False)
for t in ax.texts: t.set_text(t.get_text() + " %")
g.set(xlabel="", ylabel="")
for i in range(year_percent_pivot.shape[1]+1):
    ax.axvline(i, color='white', lw=10)
In [43]:
In [44]:
# Institutional collab
In [45]:
wos_univ_locations = wos_univ.merge(wos_country_types, on="Country")
wos_univ_collabs = wos_univ_locations[wos_univ_locations["Country_Type"]!="Other"][[record_col,"Country","Institution_harm","Country_Type"]].drop_duplicates()
wos_univ_collabs.sample(100)
Out[45]:
UT (Unique WOS ID) Country Institution_harm Country_Type
104534 WOS:000536637200011 United Kingdom Univ Warwick Non-EU associate
120323 WOS:000373806800006 France ENSAIT EU
41841 WOS:000542956600003 China Nanjing Univ Aeronaut & Astronaut China
100019 WOS:000459844300007 United Kingdom Univ Manchester Non-EU associate
174151 WOS:000843324300007 Ireland Trinity Coll Dublin EU
... ... ... ... ...
157638 WOS:000863147500001 Finland Univ Turku EU
71835 WOS:000798227800116 China Shanghai Jiao Tong Univ China
128870 WOS:000460118200077 Sweden Royal Inst Technol EU
37822 WOS:000517665600048 China Chinese Acad Sci China
26625 WOS:000453750400001 China Hangzhou Dianzi Univ China

100 rows × 4 columns

In [46]:
TOPN = 25


wos_univ_ch = wos_univ_collabs[wos_univ_collabs["Country_Type"]=="China"]
wos_univ_eu = wos_univ_collabs[wos_univ_collabs["Country_Type"]!="China"]

wos_univ_eu_strict = wos_univ_collabs[wos_univ_collabs["Country_Type"]=="EU"]

data_eu = (wos_univ_eu.groupby(["Country","Institution_harm","Country_Type"], as_index=False)[record_col].nunique()
           .sort_values(by=record_col,ascending=False).head(TOPN).copy())

data_eu_strict = (wos_univ_eu_strict.groupby(["Country","Institution_harm","Country_Type"], as_index=False)[record_col].nunique()
           .sort_values(by=record_col,ascending=False).head(TOPN).copy())
data_eu_strict

data_ch = (wos_univ_ch.groupby(["Country","Institution_harm","Country_Type"], as_index=False)[record_col].nunique()
           .sort_values(by=record_col,ascending=False).head(TOPN).copy())


for data in [data_eu,data_eu_strict,data_ch]:
    fig = px.bar(data, x=record_col, y="Institution_harm", color="Country_Type",
                              labels={
                     record_col: 'Number of co-publications',
                     "Institution_harm": "Institution",
                                  "Country_Type":"Country type"
                 },
                title="Most visible institutions", template='plotly')
    fig.update_layout(xaxis_tickformat='d',font_family="Montserrat",yaxis={'categoryorder':'total ascending'},
                                         width=1000, height=1000,)
    fig.update_traces(hovertemplate='%{x:d}')
    fig.add_shape(
            # Rectangle with reference to the plot
                type="rect",
                xref="paper",
                yref="paper",
                x0=0,
                y0=0,
                x1=1.0,
                y1=1.0,
                line=dict(
                    color="black",
                     width=0.5,
                 )
             )
    fig.update_yaxes(
        showgrid=True,
        ticks="outside")
    fig.update_xaxes(
        showgrid=True,
        ticks="outside")
    fig.show(config= dict(displayModeBar = False))
In [47]:
wos_univ_test = wos_univ_locations[wos_univ_locations["Country_Type"]!="Other"][[record_col,"Country","Institution","Institution_harm","Country_Type"]].drop_duplicates()
www = wos_univ_test.groupby(["Institution","Institution_harm"], as_index=False)[record_col].nunique()
www[www["Institution_harm"]=="Chinese Acad Sci"]
Out[47]:
Institution Institution_harm UT (Unique WOS ID)
16 Chinese Acad Sci Chinese Acad Sci 1
3149 Chinese Acad Sci Chinese Acad Sci 4614
3153 Chinese Acad Sci AIRCAS Chinese Acad Sci 2
3155 Chinese Acad Sci CAREERI CAS Chinese Acad Sci 1
3157 Chinese Acad Sci CASIA Chinese Acad Sci 8
3159 Chinese Acad Sci GUCAS Chinese Acad Sci 2
3160 Chinese Acad Sci IAP Chinese Acad Sci 1
3161 Chinese Acad Sci IECAS Chinese Acad Sci 2
3162 Chinese Acad Sci IME CAS Chinese Acad Sci 1
3163 Chinese Acad Sci IMECAS Chinese Acad Sci 1
3164 Chinese Acad Sci ITP CAS Chinese Acad Sci 1
3166 Chinese Acad Sci NAOC Chinese Acad Sci 1
3167 Chinese Acad Sci NAOC CAS Chinese Acad Sci 2
13501 RCEES Chinese Acad Sci Chinese Acad Sci 1
19499 ZIAT Chinese Acad Sci Chinese Acad Sci 1
In [48]:
wos_univ_ch = wos_univ_collabs[wos_univ_collabs["Country_Type"]=="China"]
wos_univ_eu = wos_univ_collabs[wos_univ_collabs["Country_Type"]!="China"]

wos_univ_dipol = wos_univ_eu.merge(wos_univ_ch, on=record_col, suffixes=('_eu', '_ch')).merge(wos[[record_col,"Domain_English","Field_English","SubField_English"]], on =record_col)
wos_univ_dipol.sample(100)
Out[48]:
UT (Unique WOS ID) Country_eu Institution_harm_eu Country_Type_eu Country_ch Institution_harm_ch Country_Type_ch Domain_English Field_English SubField_English
263074 WOS:000597493300001 United Kingdom Univ Northumbria Newcastle Non-EU associate China Nanchang Univ China Natural Sciences Chemistry Analytical Chemistry
71907 WOS:000494411700001 Germany Univ Wurzburg EU China South China Agr Univ China Economic & Social Sciences Social Sciences Information & Library Sciences
303069 WOS:000569985300066 Italy Selex EU China Wuhan Elect Informat Inst China Applied Sciences Engineering Computation Theory & Mathematics
259937 WOS:000557391000036 United Kingdom Univ Glasgow Non-EU associate China Southwest Jiaotong Univ China Natural Sciences Chemistry Analytical Chemistry
302133 WOS:000477943300012 Italy Politecn Milan EU China City Univ Hong Kong China Applied Sciences Information & Communication Technologies Artificial Intelligence & Image Processing
... ... ... ... ... ... ... ... ... ... ...
179087 WOS:000460128100005 Finland Univ Jyvaskyla EU China Capital Med Univ China Health Sciences Clinical Medicine Ophthalmology & Optometry
333080 WOS:000589420400001 Ireland Univ Coll Dublin EU China Guangxi Normal Univ China Natural Sciences Chemistry Organic Chemistry
300958 WOS:000388876400003 Finland Nokias Mobile Networks Organizat EU China Nokia Bell Labs China Applied Sciences Information & Communication Technologies Networking & Telecommunications
95342 WOS:000579154000008 United Kingdom Imperial Coll London Non-EU associate China Wuhan Polytech Univ China Health Sciences Clinical Medicine General & Internal Medicine
197767 WOS:000571399800004 Switzerland Univ Bern Non-EU associate China Shandong Univ China Natural Sciences Physics & Astronomy General Physics

100 rows × 10 columns

In [49]:
fig = px.parallel_categories(wos_univ_dipol[["Country_eu","Domain_English","Country_ch"]])
fig.show()
In [50]:
subfilter = ((wos_univ_dipol["Institution_harm_eu"].isin(data_eu["Institution_harm"]))&
             (wos_univ_dipol["Institution_harm_ch"].isin(data_ch["Institution_harm"])))

fig = px.parallel_categories(wos_univ_dipol[subfilter][["Country_eu","Domain_English","Country_ch"]])
fig.show()
In [51]:
subfilter = ((wos_univ_dipol["Institution_harm_eu"].isin(data_eu["Institution_harm"]))&
             (wos_univ_dipol["Institution_harm_ch"].isin(data_ch["Institution_harm"])))

fig = px.parallel_categories(wos_univ_dipol[subfilter][["Country_eu","Institution_harm_eu","Domain_English","Institution_harm_ch"]])
fig.show()
In [110]:
subfilter = ((wos_univ_dipol["Institution_harm_eu"].isin(data_eu_strict["Institution_harm"]))&
             (wos_univ_dipol["Institution_harm_ch"].isin(data_ch["Institution_harm"])))

fig = px.parallel_categories(wos_univ_dipol[subfilter][["Country_eu","Institution_harm_eu","Domain_English","Institution_harm_ch"]])
fig.show()
In [ ]:
TODO:
 - Focus sectors:
    -
In [ ]: