# Economic Impact # B. There are reports of international/foreign investors being deterred from South # African assets because of the effects of load shedding on businesses. How has # foreign investment changed with increases and decreases in load shedding? # How long does foreign investment take to react to changes in load shedding, if # at all? How would you predict foreign investment in the next 10 years as a # function of increased load shedding? # importing libraries import pandas as pd import numpy as np import seaborn as sns import matplotlib.pyplot as plt # plt.show() to display graph in code

# data showing net inflow and net outflow of foreign direct invsetment in South Africa from 2009-2014 (before load shedding) investment1 = pd.read_csv('/datasets/investment/investment1.csv') # from 2015-2020 (during load shedding) investment2 = pd.read_csv('/datasets/investment/investment2.csv')

# renaming columns for investment1 colnames1=[2009,2010,2011,2012,2013,2014] oldcolnames1 = ["2009 [YR2009]", "2010 [YR2010]", "2011 [YR2011]", "2012 [YR2012]", "2013 [YR2013]", "2014 [YR2014]"] investment1 = pd.read_csv("/datasets/investment/investment1.csv", usecols=oldcolnames1) investment1.columns = colnames1

# dropping rows 2 to 6 (empty rows) investment1 = investment1.drop(index=investment1.index[2],axis=0)

investment1 = investment1.drop([3,4,5,6])

# renaming rows for investment1 investment1.index = ['FDI net inflow','FDI net outflow']

# transposing the dataset for easier use investment1 = investment1.transpose() investment1

# renaming columns for investment2 colnames2=[2015,2016,2017,2018,2019,2020] oldcolnames2 = ["2015 [YR2015]", "2016 [YR2016]", "2017 [YR2017]", "2018 [YR2018]", "2019 [YR2019]", "2020 [YR2020]"] investment2 = pd.read_csv('/datasets/investment/investment2.csv',usecols=oldcolnames2) investment2.columns=colnames2

# dropping rows 2 to 6 (empty rows) investment2 = investment2.drop([2,6])

investment2 = investment2.drop([3,4,5])

# rename rows investment2.index = ['FDI net inflow','FDI net outflow']

# transposing the dataset for easier use investment2 = investment2.transpose()

# dropping 2020 because it is empty investment2 = investment2.drop([2020])

# converting the data to floating point numbers investment2['FDI net inflow'] = investment2['FDI net inflow'].astype(float) investment2['FDI net outflow'] = investment2['FDI net outflow'].astype(float) investment2

# dataset for foreign direct investment inflow in US dollars !ls /datasets/investmentdollars investmentdollars = pd.read_csv('/datasets/investmentdollars/investmentdollars.csv')

colnames3=[2009,2010,2011,2012,2013,2014,2015,2016,2017,2018,2019] oldcolnames3 = ["2009 [YR2009]", "2010 [YR2010]", "2011 [YR2011]", "2012 [YR2012]", "2013 [YR2013]", "2014 [YR2014]", "2015 [YR2015]", "2016 [YR2016]", "2017 [YR2017]", "2018 [YR2018]", "2019 [YR2019]"] investmentdollars = pd.read_csv("/datasets/investmentdollars/investmentdollars.csv", usecols=oldcolnames3) investmentdollars.columns = colnames3

# remove filler rows investmentdollars = investmentdollars.drop([1,2,3,4,5]) investmentdollars

investmentdollars = investmentdollars.transpose()

# adding in dataset for loadshedding history loadinghist = pd.read_csv('/datasets/loadinghist/EskomSePush Loadshedding History - EskomSePush_stages.csv') loadinghist

# create a dataset with both investment3 = pd.concat([investment1, investment2])

plt.figure(figsize=(10,5)) sns.lineplot(data=investment3, dashes=False) plt.xlabel('Year') plt.title('Direct Foreign Investment in South Africa from 2009-2019') plt.show()

investmentplot = investmentdollars.plot() investmentplot.set_xlabel("Year") investmentplot.set_ylabel("FDI in US dollars (millions)") investmentplot.set_title('Foreign Investment in South Africa')

loadinghist1 = loadinghist[loadinghist['stage'] > 0]

plt.figure(figsize=(10,5)) sns.countplot(x='stage',data=loadinghist1, palette='coolwarm') plt.xlabel('Stage') plt.title('Loadshedding From 2015-2020') plt.show()

loadinghist1['stage'].value_counts()

# convert to DateTime loadinghist1['created_at'] = pd.to_datetime(loadinghist1['created_at'])

# create new columns for hour, month, day, and year. then convert numerical days to string loadinghist1['Hour'] = loadinghist1['created_at'].apply(lambda x: x.hour) loadinghist1['Month'] = loadinghist1['created_at'].apply(lambda x: x.month) loadinghist1['Day'] = loadinghist1['created_at'].apply(lambda x: x.dayofweek) loadinghist1['Year'] = loadinghist1['created_at'].apply(lambda x: x.year) dmap = {0:'Mon',1:'Tue',2:'Wed',3:'Thu',4:'Fri',5:'Sat',6:'Sun'} loadinghist1['Day'] = loadinghist1['Day'].apply(lambda x: dmap[x])

sns.countplot(x = 'Day', data = loadinghist1) plt.title('Frequency of Loadsheddings Per Day') plt.show()

sns.countplot(x = 'Year', data = loadinghist1) plt.title('Frequency of Loadsheddings Per Year') plt.show()

loadinghist1['Year'].value_counts()

sns.countplot(x = 'Month', data = loadinghist1) plt.title('Frequecncy of Loadsheddings Per Month') plt.show()

g = sns.FacetGrid(loadinghist1, col='Year') g.map_dataframe(sns.histplot, x="Month") plt.show()

sns.countplot(x = 'Hour', data = loadinghist1) plt.title('Frequency of Loadsheddings per Hour') plt.show()

# creating a dataframe with the number of loadsheddings per year loadinghist_year = pd.DataFrame(loadinghist1['Year'].value_counts()) loadinghist_year.columns = ['Num Loads']

# adding new dataframe to original one investmentwithloads = pd.concat([investment3, loadinghist_year], axis=1) # convert Nan values to 0 investmentwithloads = investmentwithloads.fillna(0) # drop 2020 row because there is no data on FDI investmentwithloads = investmentwithloads.drop([2020])

investmentwithloads.corr()

testdf = pd.concat([investment3, loadinghist_year], axis=1) testdf = testdf.fillna(0) testdf = testdf.drop([2009,2010,2011,2012,2013,2014,2015,2016,2020])

testdf.corr()