How best can resources be allocated in order to reduce unemployment in South Africa in a post Covid-era?¶
In [29]:
from __future__ import print_function # For Python 2 / 3 compatability
%matplotlib inline
import numpy as np
import pandas as pd
import seaborn as sns
import chardet
import math
import matplotlib.pyplot as plt
#np.random.seed(0)
#%pip install cchardet
#import cchardet as chardet
saved excel spreadsheet as a CSV¶
file = open(r'C:\Users\user\Downloads\ethekwini_skills.xls',encoding='utf-8') file
In [30]:
data=pd.read_csv(file)
data.head()
In [31]:
# saved excel spreadsheet as a CSV
file = open(r'C:\Users\user\Downloads\ethekwini_skill.csv')
file
Out[31]:
In [32]:
data=pd.read_csv(file)
data.head()
Out[32]:
In [33]:
data.tail()
Out[33]:
In [34]:
new_header = data.iloc[0] #grab the first row for the header
data = data[1:] #take the data less the header row. Essentially taking all the data going downwards
data.columns = new_header #set the header row as the df header
data.head()
Out[34]:
In [35]:
data.shape
Out[35]:
In [36]:
data.info()
In [37]:
sns.heatmap(data.isnull(),yticklabels=False,cmap='viridis')
Out[37]:
In [38]:
data.columns
Out[38]:
In [39]:
data.nunique()
Out[39]:
In [40]:
data['Gender'].unique()
Out[40]:
In [41]:
data['Disability'].unique()
Out[41]:
In [42]:
data['Skill Sector'].unique()
Out[42]:
In [43]:
for col_name in data.columns:
if data[col_name].dtypes == 'object':
unique_cat = len(data[col_name].unique())
print("Feature '{col_name}' has {unique_cat} unique categories".format(col_name=col_name,unique_cat=unique_cat))
In [44]:
# first three for incomplete data and the next two for irrelevance/redundancy
ndata = data.drop(['Skill Description','Skill Experience Duration','Informal Training Description','Suburb','Informal Experience Duration'], axis = 1) # Policy can be made with regards to wards and not suburbs
ndata.head()
Out[44]:
In [45]:
sns.heatmap(ndata.isnull(),yticklabels=False,cmap='viridis')
Out[45]:
In [46]:
dataTypeSeries = ndata.dtypes
print('Data type of each column of Dataframe :')
print(dataTypeSeries)
In [47]:
ndata['Age'] = ndata['Age'].astype(int)
print (ndata['Age'].dtypes)
In [48]:
ndata['Age'].plot.hist()
Out[48]:
In [49]:
sns.distplot(ndata['Age'])
Out[49]:
In [50]:
ndata['Age'].describe()
Out[50]:
In [51]:
ndata['Age'] = pd.to_numeric(ndata['Age'])
ndata['Age'] = np.where(ndata['Age'] > 15, '15-20', ndata['Age'])
ndata.head()
Out[51]:
In [52]:
nndata = ndata.loc[ndata['Ward No']=='28']
nndata.head(5)
Out[52]:
In [53]:
nndata.shape
Out[53]:
In [54]:
print (ndata['Age'].dtypes)
In [55]:
plt.figure(figsize=(16, 16))
ax1 = sns.countplot(x="Skill Sector",data=nndata)
ax1.set_xticklabels(ax1.get_xticklabels(), rotation=40, ha="right")
plt.tight_layout()
plt.show()
In [56]:
plt.figure(figsize=(16, 16))
ax2 = sns.countplot(x="Skill Sector",data=ndata)
ax2.set_xticklabels(ax2.get_xticklabels(), rotation=40, ha="right")
plt.tight_layout()
plt.show()
In [58]:
plt.figure(figsize=(8, 8))
ax4 = sns.countplot(x='Qualification', data=ndata)
ax4.set_xticklabels(ax4.get_xticklabels(), rotation=40, ha="right")
plt.tight_layout()
plt.show()
In [59]:
sns.countplot(x="Disability",data=ndata)
Out[59]:
In [60]:
ndata['Disability'].value_counts()
Out[60]:
In [61]:
ndata['Gender'].value_counts() #less than 1% unspecified and 44% male
Out[61]:
In [62]:
genders = pd.get_dummies(ndata['Gender'],drop_first=True)
license = pd.get_dummies(ndata['Drivers Licence'],drop_first=True)
disability = pd.get_dummies(ndata['Disability'],drop_first=True)
In [63]:
genders.head(5)
Out[63]:
In [64]:
ndata.drop(['Gender'],axis=1,inplace=True)
In [65]:
ndata.drop(['Disability'],axis=1,inplace=True)
In [66]:
ndata.drop(['Drivers Licence'],axis=1,inplace=True)
In [67]:
ndata.head()
Out[67]:
In [68]:
ndata=pd.concat([ndata,genders,license,disability],axis=1)
In [69]:
ndata
Out[69]:
In [70]:
# Qualitative study by Moodley,S 2017 GIBS for MBA
gibs_study = {"Formal,Informal & Non-formal": 5, "Formal & Non-formal": 4, "Formal & Informal": 2, "Formal": 1, "Informal": 1, "Non-formal": 0}
plt.pie([float(v) for v in gibs_study.values()], labels=[k for k in gibs_study.keys()],
autopct='%1.0f%%', pctdistance=0.8, labeldistance=1.2)
Out[70]:
In [71]:
dataz = pd.read_csv(r'C:\Users\user\Downloads\ethekwini-business-license-data-in-all-regions.csv')
dataz.head(10)
Out[71]:
In [72]:
dataz.drop(['Proprietor', 'BusinessName','RefNo','PhysicalAddress', 'Telephone','PostalAddress','LicenseSubType','CurrentDate'],axis=1,inplace=True)
dataz.head()
Out[72]:
In [73]:
dataz.drop(['Employer', 'AnnualNotification','LicenseIssueDate','NotificationUpdateDate', 'Unnamed: 13','Unnamed: 14'],axis=1,inplace=True)
In [74]:
dataz.head()
Out[74]:
In [75]:
dataz.tail()
Out[75]:
In [76]:
dataz.info()
In [77]:
dataz.nunique()
Out[77]:
In [79]:
plt.figure(figsize=(8, 8))
ax3 = sns.countplot(x='LicenseType',data=dataz)
ax3.set_xticklabels(ax3.get_xticklabels(), rotation=40, ha="right")
plt.tight_layout()
plt.show()
In [80]:
dataz.head(50)
Out[80]:
In [81]:
datas = pd.read_csv(r'C:\Users\user\Downloads\unemploymentrates.csv')
datas.head(10)
Out[81]:
In [82]:
datas = datas.iloc[:-1] #take the data less the header row. Essentially taking all the data going downwards
datas.head(10)
Out[82]:
In [83]:
corelation = datas.corr()
sns.heatmap(corelation, xticklabels=corelation.columns, yticklabels=corelation,annot=True)
Out[83]:
In [ ]:
from google
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]: